5 This is the list of official releases for libvirt, along with an overview of
6 the changes introduced by each of them.
8 For a more fine-grained view, use the `git log`_.
16 * **Removed features**
20 * qemu: Support for ras feature for virt machine type
22 It is now possible to set on/off ``ras`` feature in the domain XML for virt
23 (Arm) machine type as ``<ras state='on'/>``.
27 Libvirt now installs a binary helper that allows connecting to QEMU domains
28 via SSH using the following scheme: ``ssh user@qemu/virtualMachine``.
30 * qemu: Support for ``virtio`` sound model
32 Sound devices can now be configured to use the virtio model with
33 ``<sound model='virtio'/>``. This model is available from QEMU 8.2.0
40 * virsh/virt-admin: Fix ``--help`` option for all commands
42 A bug introduced in `v10.3.0 (2024-05-02)`_ caused that the attempt to print
43 help for any command by using the ``--help`` option in ``virsh`` and
44 ``virt-admin`` would print::
47 error: command 'list' doesn't support option --help
49 instead of the help output. A workaround for the affected version is to use
54 * qemu: Fix ``virsh save`` and migration when storage in question is root_squashed NFS
56 Attempting to save a VM to a root_squash NFS mount or migrating with disks
57 hosted on such mount could, in some scenarios, result in error stating::
61 The bug was introduced in `v10.1.0 (2024-03-01)`_.
69 * qemu: Proper support for USB network device
71 USB address is now automatically assigned to USB network devices thus they
72 can be used without manual configuration.
74 * conf: Introduce memReserve attribute to <controller/>
76 Some PCI devices have large non-prefetchable memory. This can be a problem
77 in case when such device needs to be hotplugged as the firmware can't
78 foresee such situation. The user thus can override the value calculated at
79 start to accomodate for such devices.
83 * Improve validation of USB devices
85 Certain USB device types ('sound', 'fs', 'chr', 'ccid' and 'net') were not
86 properly handled in the check whether the VM config supports USB and thus
87 would result in poor error messages.
89 * virsh: Fix behaviour of ``--name`` and ``--parent`` used together when listing checkpoint and snapshots
91 The ``checkpoint-list`` and ``snapshot-list`` commands would ignore the
92 ``--name`` option to print only the name when used with ``--parent``.
94 * Extend libvirt-guests to shutdown only persistent VMs
96 Users can now choose to shutdown only persistent VMs when the host is being
101 * qemu: Fix migration with custom XML
103 Libvirt 10.2.0 would sometimes complain about incompatible CPU definition
104 when trying to migrate or save a domain and passing a custom XML even
105 though such XML was properly generated as migratable. Hitting this bug
106 depends on the guest CPU definition and the host on which a particular
109 * qemu: Fix TLS hostname verification failure in certain non-shared storage migration scenarios
111 In certain scenarios (parallel migration, newly also post-copy migration)
112 libvirt would wrongly pass an empty hostname to QEMU to be used for TLS
113 certificate hostname validation, which would result into failure of the
114 non-shared storage migration step::
116 error: internal error: unable to execute QEMU command 'blockdev-add': Certificate does not match the hostname
118 * Create OVS ports as transient
120 Libvirt now creates OVS ports as transient which prevents them from
121 reappearing or going stale on sudden reboots.
123 * Clear OVS QoS settings when domain shuts down
125 Libvirt now clears QoS settings on domain shutdown, so they no longer pile
134 * ch: Basic save and restore support for ch driver
136 The ch driver now supports basic save and restore operations. This is
137 functional on domains without any network, host device config defined.
138 The ``path`` parameter for save and restore should be a directory.
140 * qemu: Support for driver type ``mtp`` in ``<filesystem/>`` devices
142 The ``mtp`` driver type exposes the ``usb-mtp`` device in QEMU. The
143 guest can access files on this driver through the Media Transfer
146 * qemu: Added support for the loongarch64 architecture
148 It is now possible for libvirt to run loongarch64 guests, including on
149 other architectures via TCG. For the best results, it is recommended to
150 use the upcoming QEMU 9.0.0 release together with the development version
153 * qemu: Introduce virDomainGraphicsReload API
155 Reloading the graphics display is now supported for QEMU guests using
156 VNC. This is useful to make QEMU reload the TLS certificates without
157 restarting the guest. Available via the ``virDomainGraphicsReload`` API
158 and the ``domdisplay-reload`` virsh command.
162 * qemu: Fix migration from libvirt older than 9.10.0 when vmx is enabled
164 A domain with vmx feature enabled (which may be even done automatically
165 with ``mode='host-model'``) started by libvirt 9.9.0 or older cannot be
166 migrated to libvirt 9.10.0, 10.0.0, and 10.1.0 as the target host would
167 complain about a lot of extra ``vmx-*`` features. Migration of similar
168 domains started by the affected releases to libvirt 9.9.0 and older
169 does not work either. Since libvirt 10.2.0 migration works again with
170 libvirt 9.9.0 and older in both directions. Migration from the affected
171 releases to 10.2.0 works as well, but the other direction remains broken
172 unless the fix is backported.
174 * node_device: Don't report spurious errors from PCI VPD parsing
176 In last release the PCI Vital Product Data parser was enhanced to report
177 errors but that effort failed as some kernels have the file but don't allow
178 reading it causing logs to be spammed with::
180 libvirtd[21055]: operation failed: failed to read the PCI VPD data
182 Since the data is used only in the node device XML and errors are ignored if
183 the parsing failed, this release removes all the error reporting.
185 * qemu: set correct SELinux label for unprivileged virtiofsd
187 It is now possible to use virtiofsd-based ``<filesystem>`` shares even
188 if the guest is confined using SELinux.
190 * qemu: fix a crash on unprivileged virtiofsd hotplug
192 Hotplugging virtiofsd-based filesystems works now.
194 * virt-admin: Fix segfault when libvirtd dies
196 ``virt-admin`` no longer crashes when ``libvirtd`` unexpectedly closes
205 * ``CVE-2024-1441``: Fix off-by-one error leading to a crash
207 In **libvirt-1.0.0** there were couple of interface listing APIs
208 introduced which had an off-by-one error. That error could lead to a
209 very rare crash if an array was passed to those functions which did
210 not fit all the interfaces.
212 In **libvirt-5.10** a check for non-NULL arrays has been adjusted to
213 allow for NULL arrays with size 0 instead of rejecting all NULL
214 arrays. However that made the above issue significantly worse since
215 that off-by-one error now did not write beyond an array, but
216 dereferenced said NULL pointer making the crash certain in a
217 specific scenario in which a NULL array of size 0 was passed to the
218 aforementioned functions.
222 * nodedev: Support updating mdevs
224 The node device driver has been extended to allow updating mediated node
225 devices. Options are available to target the update against the persistent,
226 active or both configurations of a mediated device.
227 **Note:** The support is only available with at least mdevctl v1.3.0 installed.
229 * qemu: Add support for /dev/userfaultfd
231 On hosts with new enough kernel which supports /dev/userfaultfd libvirt will
232 now automatically grant QEMU access to this device. It's no longer needed to
233 set vm.unprivileged_userfaultfd sysctl.
235 * qemu: Support clusters in CPU topology
237 It is now possible to configure the guest CPU topology to use clusters.
238 Additionally, if CPU clusters are present in the host topology, they will
239 be reported as part of the capabilities XML.
241 * network: Make virtual domains resolvable from the host
243 When starting a virtual network with a new ``register='yes'`` attribute
244 in the ``<domain>`` element, libvirt will configure ``systemd-resolved``
245 to resolve names of the connected guests using the name server started
248 * qemu: Introduce dynamicMemslots attribute for virtio-mem
250 QEMU now allows setting ``.dynamic-memslots`` attribute for virtio-mem-pci
251 devices. When turned on, it allows memory exposed to guest to be split into
252 multiple memory slots and thus smaller memory footprint (see the original
253 commit for detailed explanation).
257 * nodedev: Add ability to update persistent mediated devices by defining them
259 Existing persistent mediated devices can now also be updated by
260 ``virNodeDeviceDefineXML()`` as long as parent and UUID remain unchanged.
262 * ch: Enable ``ethernet`` interface mode support
264 ``<interface type='ethernet'/>`` can now be used for CH domains.
266 * viraccessdriverpolkit: Add missing vtpm case
268 Secrets with ``<usage type='vtpm'>`` were left unable to be checked for in
269 the access driver, i.e. in ACL rules. Missing code was provided.
271 * virt-admin: Notify users to use explicit URI if connection fails
273 ``virt-admin`` doesn't try to guess the URI of the daemon to manage so a
274 failure to connect may be confusing for users if modular daemons are used.
275 Add a hint to use the URI of the dameon to manage.
279 * qemu_process: Skip over non-virtio non-TAP NIC models when refreshing rx-filter
281 If ``trustGuestRxFilters`` is enabled for a vNIC that doesn't support it,
282 libvirt may throw an error when such domain is being started, loaded from a
283 saved state, migrated, etc. These errors are now silenced, but make sure to
284 fix such configurations (after previous release it is even possible to
285 change ``trustGuestRxFilters`` value on live domains via
286 ``virDomainUpdateDeviceFlags()`` or ``virsh device-update``).
288 * domain: Fix check for overlapping ``<memory/>`` devices
290 A bug was identified which caused libvirt to report two NVDIMMs as
291 overlapping even though they weren't. This now fixed.
293 * vmx: Accept empty fileName for cdrom-image
295 Turns out, ``fileName`` attribute (which contains path to CDROM image) can
296 be set to an empty string (``""``) to denote a state in which the CDROM has
297 no medium in it. Libvirt used to reject such configuration file, but not
300 * qemu_hotplug: Don't lose 'created' flag in qemuDomainChangeNet()
302 When starting a domain, libvirt tracks what resources it created for it and
303 which were pre-existing and uses this information to preserve pre-existing
304 resources when cleaning up after said domain is shut off. But for macvtaps
305 this information was lost after the macvtap device was changed (e.g. via
306 ``virsh update-device``).
308 * Fix virStream hole handling
310 When a client sent multiple holes into a virStream it may have caused
311 daemon hangup as the daemon stopped processing RPC from the client
312 temporarily. This is now fixed.
314 * nodedev: Don't generate broken XML with certain hardware
316 A broken node device XML would be generated in a rare case when a hardware
317 device had certain characters in the VPD fields.
319 * qemu: Fix reservation of manually specified port for disk migration
321 A manually specified port would not be relased after disk migration making
322 it impossible to use it again.
330 * qemu: Enable ``postcopy-preempt`` migration capability
332 Post-copy migrations are now started with ``postcopy-preempt``
333 capability enabled as long as it is supported by both sides of migration.
334 This should enable faster migration of memory pages that the destination
335 tries to read before they are migrated from the source.
337 * qemu: Add support for mapping iothreads to virtqueues of ``virtio-blk`` devices
339 QEMU added the possibility to map multiple ``iothreads`` to a single
340 ``virtio-blk`` device and map them even to specific virtqueues. Libvirt
341 adds a ``<iothreads>`` subelement of the ``<disk> <driver>`` element that
342 users can use to configure the mapping.
344 * qemu: Allow automatic resize of block-device-backed disk to full size of the device
346 The new flag ``VIR_DOMAIN_BLOCK_RESIZE_CAPACITY`` for
347 ``virDomainBlockResize`` allows resizing a block-device backed ``raw`` disk
348 of a VM without the need to specify the full size of the block device.
350 * qemu: automatic selection/binding of VFIO variant drivers
352 When a device is assigned to a guest using VFIO with ``<hostdev
353 managed='yes'>``, libvirt will now search the running kernel's
354 modules.alias file for the most specific match to that device for
355 a VFIO driver, and bind that driver to the device rather than
356 vfio-pci. A specific driver can also be forced, using the
357 ``<driver model='plugh'/>`` attribute.
359 * qemu: add runtime configuration option for nbdkit
361 Since the new nbdkit support requires a recent selinux policy that is not
362 widely available yet, it is now possible to build libvirt with nbdkit
363 support for remote disks but disabled at runtime. This behavior is
364 controlled via the storage_use_nbdkit option of the qemu driver
365 configuration file. The option will default to being disabled, but this may
366 change in a future release and can be customized with the
367 nbdkit_config_default build option.
369 * qemu: add ID mapping support for virtiofsd
371 New ``<idmap>`` element was added for virtiofsd-based ``<filesystem>``
372 devices. It can be used to set up UID and GID mapping between host
373 and guest, making running virtiofsd unprivileged much more useful.
377 * qemu: Improve migration XML use when persisting VM on destination
379 When migrating a VM with a custom migration XML, use it as a base for
380 persisting it on the destination as users could have changed non-ABI
381 breaking facts which would prevent subsequent start if the old XML were used.
383 * qemu: Simplify non-shared storage migration to ``raw`` block devices
385 The phase of copying storage during migration without shared storage
386 requires that both the source and destination image are identical in size.
387 This may not be possible if the destination is backed by a block device
388 and the source image size is not a multiple of the block device block size.
390 Libvirt aleviates this by automatically adding a ``<slice>`` to match the
391 size of the source image rather than failing the migration.
393 * test driver: Support for hotplug/hotunplug of PCI devices
395 The test driver now supports basic hotplug and hotunplug of PCI devices.
397 * qemu: allow virtiofsd to run unprivileged
399 Nowadays virtiofsd no longer requires to run with root privileges, so the
400 restriction to always run as root is now removed from libvirt too.
404 * qemu: Various migration bug fixes and debuggability improvement
406 This release fixes multiple bugs in virsh and libvirt in handling of
407 migration arguments and XMLs and modifies error reporting for better
410 * conf: Restore setting default bus for input devices
412 Because of a regression, starting from 9.3.0 libvirt did not autofill bus
413 for input devices. With this release the regression was identified and
416 * qemu: Relax check for memory device coldplug
418 Because of a check that was too aggressive, a virtio-mem memory device
419 could not be cold plugged. This is now fixed.
421 * qemu: Be less aggressive when dropping channel source paths
423 Another regression is resolved, (introduced in 9.7.0) when libvirt was too
424 aggressive when dropping parsed paths for <channel/> sources
426 * qemuDomainChangeNet: Reflect trustGuestRxFilters change
428 On device-update, when a user requested change of trustGuestRxFilters for a
429 domain's <interface/> libvirt did nothing. It did not throw an error nor
430 did it reflect the change. Starting with this release, the change is
439 * Introduce pipewire audio backend
441 The QEMU hypervisor driver now allows setting ``pipewire`` backend for
446 * Adapt to qemu's use of protocol drivers in QCOW2 'backing file format' field
448 QEMU allows creating images where the 'backing file format' is actually a
449 protocol name such as 'file'/'host_device'/'nbd'/etc.. Adapt libvirt to
450 properly handle such images and don't assume automatic format probing is
451 necessary, which is in many cases forbidden due to security implications.
455 * qemu: Fix setup of images on hotplug of disk
457 Internal image metadata was not setup correctly which could cause some disk
458 hotplug configurations (namely those including backing images) to fail.
460 * qemu: Fix qemu crash when reverting an internal snapshot
462 Libvirt attempted to start qemu with wrong arguments when attempting to
463 revert to an internal snapshot causing qemu to crash.
465 * qemu: Fix hotplug of empty cdrom
467 Empty cdrom drive couldn't be hotplugged as libvirt wanted to setup the
468 storage backing it unconditionally.
476 * QEMU: implement reverting external snapshots
478 Reverting external snapshots is now possible using the existing API
479 ``virDomainSnapshotRevert()``. Management application can check host
480 capabilities for ``<externalSnapshot/>`` element within the list of
481 guest features to see if the current libvirt supports both deleting
482 and reverting external snapshots.
484 * virsh: add ``console --resume`` support
486 The ``virsh console`` subcommand now accepts a ``--resume`` option. This
487 will resume a paused guest after connecting to the console.
491 * virsh: Improve ``virsh start --console`` behavior
493 The ``virsh start --console`` now tries to connect to the guest console
494 before starting the vCPUs.
496 * virsh: Improve ``virsh create --console`` behavior
498 The ``virsh create --console`` now tries to connect to the guest console
499 before starting the vCPUs.
507 * network: New metadata change event
509 The network object now has a new event ID ``VIR_NETWORK_EVENT_ID_METADATA_CHANGE``
510 that can be used to get notifications upon changes in any of ``<title>``,
511 ``<description>`` or ``<metadata>``.
513 * qemu: Add support for vDPA block devices
515 With a new enough version of qemu, libvirt will allow you to assign vDPA block
516 devices to a domain. This is configured with::
518 <disk type='vhostvdpa'>
519 <source dev='/dev/vhost-vdpa-0'>
524 * qemu: add nbdkit backend for network disks
526 Up until now, libvirt supported network disks (http, ftp, ssh) by passing
527 the URL to qemu and having the appropriate qemu block drivers handle the
528 disk I/O. However, by handling the network I/O outside of the qemu process,
529 we get several advantages, such as reduced attack surface and improved
530 stability of qemu. Therefore, when available, libvirt will use nbdkit as a
531 backend for these network disks and export an NBD disk to qemu.
533 * virnetdevopenvswitch: Propagate OVS error messages
535 When configuring OVS interfaces/bridges libvirt used to report its own
536 error messages instead of passing (more accurate) error messages from
537 `ovs-vsctl`. This is now changed.
539 * Various virtio-mem/virtio-pmem fixes
541 Now libvirt validates more values of virtio-mem and virtio-pmem devices,
542 e.g. overlapping memory addresses or alignment.
550 * qemu: basic support for use of "VFIO variant" drivers
552 A VFIO variant driver is a device-specific driver that can
553 be used in place of the generic vfio-pci driver, and provides
554 extra functionality to support things like live migration of
555 guests with vfio-assigned devices. It can currently be used by:
557 1) setting ``managed='no'`` in the XML configuration for the device
558 2) pre-binding the variant driver using the ``--driver`` option of
559 ``virsh nodedev-detach``.
561 * network: Support for ``<title>`` and ``<description>`` fields in Network XML
563 The network object adds two more user defined metadata fields ``<title>``
564 and ``<description>``.
565 Two new APIs ``virNetworkGetMetadata()`` and ``virNetworkSetMetadata()`` can be
566 used to view and modify the above including the existing ``<metadata>`` field.
568 virsh adds two new commands ``net-desc`` and ``net-metadata`` to view/modify the same.
569 ``net-list`` adds a new option ``--title`` that prints the content of ``<title>``
570 in an extra column within the default ``--table`` output.
574 * qemu: Various fixes to firmware selection
576 The changes made to firmware selection in libvirt 9.2.0 have unfortunately
577 introduced a number of regressions. All known issues in this area have now
586 * ``CVE-2023-3750``: Fix race condition in storage driver leading to a crash
588 In **libvirt-8.3** a bug was introduced which in rare cases could cause
589 ``libvirtd`` or ``virtstoraged`` to crash if multiple clients attempted to
590 look up a storage volume by key, path or target path, while other clients
591 attempted to access something from the same storage pool.
595 * apparmor: All profiles and abstractions now support local overrides
597 This has long been the case for the ``virt-aa-helper`` profile, but has
598 now been extended to all other profiles and abstractions. The mechanism
599 used is the standard AppArmor 3.x one, where the contents of ``foo`` and
600 ``abstractions/foo`` can be overridden by creating ``local/foo`` and
601 ``abstractions/foo.d`` respectively.
603 * qemu: Support ``removable`` attribute for scsi disk
605 Now the scsi disk device (``/disk@device='disk'`` and
606 ``/disk/target@bus='scsi'``) supports the ``removable`` attribute at
607 ``/disk/target@removable```.
609 * qemu: Add NUMA node automatically for memory hotplug
611 Users no longer need to specify guest NUMA node in the domain XML when
612 enabling memory hotplug, libvirt automatically adds one when it is missing.
614 * qemu: Consider ``BeeGFS`` as a shared filesystem
616 Allow migration with non-shared storage for VMs accessing storage via
621 * qemu: Adapt to new way of specifying PC speaker
623 PC speaker is now usable again with newer QEMU since the change of how it
624 is specified on the command line.
626 * qemu_tpm: Try harder to create emulator state
628 Libvirt no longer considers empty directory valid SWTPM state and setup is
629 now run properly in such case.
637 * qemu: Allow configuring the ``discard-no-unref`` feature of ``qcow2`` driver
639 The new ``discard_no_unref`` attribute of the ``disk`` ``driver`` element
640 controls whether the ``qcow2`` driver in qemu unrefs clusters inside the
641 image on discard requests. Disabling cluster unrefing decreases fragmentation
646 * qemu: Include maximum physical address size in baseline CPU
648 When computing a baseline CPU definition for a set of hosts, we need to
649 include maximum physical address size in the result to make sure it is
650 compatible with all hosts even if their supported physical address sizes
653 * conf: Properly handle slots for non-DIMM ``<memory>`` devices
655 Memory devices such as ``virtio-mem`` don't need a memory slot as they are
656 PCI devices. ``libvirt`` now properly accounts the memory slots for such
657 devices as well as specifying the ``slots`` attribute of the ``<maxMemory>``
658 element is no longer needed unless DIMM-like devices are to be used.
660 * ``passt`` log and port forwarding improvements
662 Libvirt now ensures that the ``passt`` helper process can access the
663 configured log file even when it's placed in a directory without permissions.
665 The ``<portForward>`` element of a passt-backed interface can now omit the
666 ``address`` attribute as it's enough to specify a ``dev``.
670 * lxc: Allow seeking in ``/proc/meminfo`` to resove failure with new ``procps`` package
672 New version of the ``free`` command from ``procps`` package seeks into the
673 ``/proc/meminfo`` file, which was not supported by the instance of the file
674 exposed via LXC causing a failure.
676 * qemu: Fix rare race-condition when detaching a device
678 The device removal handler callback function didn't re-check the state of
679 the unplug operation after a timeout, which could rarely cause that the
680 device was removed from the VM but not the definition.
682 * qemu: Fix NUMA memory allocation logic
684 QEMU allocates memory via the emulator thread thus that has to be allowed
685 to access all configured NUMA nodes of the VM rather than just the one where
686 it's supposed to be pinned.
688 * qemu: Fix setup of ``hostdev`` backed ``<interface>``
690 The proper steps to initialize the host device were skipped for interfaces
691 due to a logic bug preventing start of VM which used them.
699 * qemu: Support compression for parallel migration
701 QEMU supports parallel migration to be compressed using either zstd or zlib.
703 * cpu_map: Add SapphireRapids cpu model
705 This model is introduced since QEMU 8.0.
709 * Adapt to musl-1.2.4
711 The latest version of musl stopped declaring some symbols that libvirt's
712 test suite used (for redirecting ``stat()`` family of functions), leaving
713 the tests broken. This is now fixed and the test suite works even with the
714 latest version of musl.
716 * conf: Introduce ``<address/>`` for virtio-mem and virtio-pmem
718 To ensure guest ABI stability, libvirt persists address for memory devices,
719 now including ``virtio-mem`` and ``virtio-pmem``. The address can be also
724 * qemu: Account for NVMe disks when calculating memlock limit on hotplug
726 When no ``<hard_limit/>`` is set, libvirt still tries to guess a sensible
727 limit for memlock for domains. But this limit was not calculated properly
728 on a hotplug of ``<disk type='nvme'/>``.
730 * numa: Deny other memory modes than ``restrictive``` if a memnode is ``restrictive``
732 Due to a missing check it was possible to define a domain with incorrect
733 ``<numatune/>``. For instance it was possible to have a ``<memnode
734 mode="restrictive"/>`` and ``<memory/>`` of a different mode. This is now
735 forbidden and if either all ``<memnode/>``-s and ``<memory/>`` have to have
736 ``restrictive`` mode, or none.
738 * qemu: Start emulator thread with more generous ``cpuset.mems``
740 To ensure memory is allocated only from configured NUMA nodes, libvirt sets
741 up cpuset CGgroup controller, even before QEMU is executed. But this may
742 prevent QEMU from setting affinity of threads that allocate memory. Since
743 these threads are spawned from the emulator thread, the initial set up must
744 be more generous and include union of all host NUMA nodes that are allowed
745 in the domain definition. Once QEMU has allocated all its memory, the
746 emulator thread is restricted further, as it otherwise would be.
754 * qemu: Introduce support for ``igb`` network interface model
756 ``igb`` is a successor to the ``e1000e`` network device using PCIe interface.
757 It was introduced in QEMU 8.0
759 * qemu: Improve handling of maximum physical address configuration
763 * qemu: Change default machine type for ARM and RISC-V
765 ARM and RISC-V architectures now use the ``virt`` machine type by default.
766 The previous defaults were nearly unusable and had to be overridden in most
769 * Improve translatable strings format substitutions
771 All translatable error messages with substitution strings were converted to
772 use positional modifiers to allow translators to shuffle around words in
773 the translation. The translations in Weblate were also updated to match.
775 * qemu: Improve validation of ``watchdog`` devices
777 Certain invalid configurations of ``watchdog`` device are now properly
780 - hotplug of always-present platform watchdogs is forbidden
781 - ``iTCO`` watchdog can be configured only once
782 - ``ib700`` watchdog is allowed only on ``i440fx`` machines
784 * Improved output of ``virt-host-validate`` on ARM
786 Our validation tool now parses the ``IORT`` data on ARM to properly detect
787 presence of SMMU and other features.
791 * qemu: Fix inactive internal snapshots of VM with UEFI firmware
793 Recent changes to UEFI firmware handling resulted into breaking support
794 for inactive internal snapshots of VMs with UEFI which historically worked.
795 (Although the intention was to disallow them together with active ones, but
796 the check did not work properly.)
798 Preserve existing functionality by allowing such snapshots explicitly.
800 * qemu: Properly configure locked memory limit for VMs with ``<disk type='nvme'``
802 The NVMe driver in qemu requires some memory to be locked. This was not
803 taken into account in the code which calculates the memory limits based
804 on devices present in the configuration
806 * Fix native build on win32
808 Various improvements to the build system now allow users to build the client
809 library of libvirt on win32 natively.
811 * qemu: Properly detect tray of hotplugged CD-ROM devices
813 Media in a CD-ROM device which was hotplugged could not be changed as the
814 presence of the tray was not detected properly on hotplug.
822 * qemu: Add support for QCOW2 formatted firmware
824 This type of firmware can be picked up either automatically, if the
825 corresponding JSON descriptor has the highest priority, or manually by
826 using ``<loader format='qcow2'/>`` in the domain XML.
828 * qemu: Implement QEMU NBD reconnect delay attribute
830 Support the nbd reconnect-delay of QEMU. It will set the delay time for
831 reconnect after an unexpected disconnect or a serious error.
835 * qemu: Make firmware selection persistent
837 Up until now, firmware autoselection has been performed at domain startup
838 time: as a result, changes to the JSON firmware descriptors present on the
839 system could have translated to a different firmware being chosen for
840 subsequent startups of the same domain, potentially rendering it unbootable
841 or lowering the security guarantees. Firmware selection now happens once,
842 when the domain is defined, and its results are stored in the domain XML
843 to be reused, unchanged, for all subsequent boots.
845 * qemu: passt now works when SELinux/AppArmor is enabled
847 In the case of SELinux, this requires passt-specific support code to be
848 present in the host policy, so it might only work with upcoming operating
849 systems and not with existing ones.
851 * xen: Support custom UEFI firmware paths
853 The Xen libxl driver now supports specifying a custom UEFI firmware path.
854 Previously the Xen default was used in all cases.
858 * qemu: Fix validation of the HPET timer
860 Due to a logic bug introduced in libvirt 9.0.0, VM configurations
861 explicitly enabling the HPET timer were rejected.
863 * qemu: Fix thread-context .host-nodes generation
865 With new enough QEMU, libvirt instructs QEMU to set affinity of memory
866 allocation threads. But this may have resulted in QEMU being unable to do
867 so, as affinity to NUMA nodes inaccessible to emulator thread might have
870 * rpc: fix typo in admin code generation
872 Fix the bug in the remote ``virt-admin`` code generator, that resulted
873 in a crash. Introduced in libvirt 9.1.0.
875 * qemu: relax shared memory check for vhostuser daemons
877 Fix hotplug of virtiofs ``filesystem`` after restarting libvirtd.
878 Before, libvirtd would incorrectly complain about missing shared
885 * **Removed features**
887 * vbox: removed support for version 5.2 and 6.0 APIs
889 Libvirt no longer supports use of VirtualBox 5.2 and 6.0 since these
890 versions reached their end of life on 2020/07.
894 * vbox: added support for version 7.0 API
896 Libvirt can now support use of the VirtualBox 7.0, This is compile tested
897 only, so we are looking for feedback from users on how well it works in
900 * qemu: Support crypto device
902 Support crypto device(virtio crypto only), also add support for QEMU with
903 backend ``builtin`` and ``lkcf``.
905 * qemu: added support for pvpanic-pci device
907 A pvpanic device can be now defined as a PCI device (the original is an ISA
908 device) with ``<panic model='pvpanic'/>``.
910 * qemu: support automatic restart of inadvertently terminated passt process
912 If the passt process that is serving as the backend of a -netdev
913 stream is terminated unexpectedly, libvirt now listens to QEMU's
914 notification of this, and starts up a new passt instance, thus
915 preserving network connectivity.
919 * RPM packaging changes
921 The ``libvirt-daemon`` subpackage is split into several new subpackages,
922 allowing installation of a modular daemon configuration without the
923 traditional monolithic libvirtd.
927 * QEMU: iTCO watchdog made operational
929 The watchdog was always included when q35 machine type was used, but needed
930 an extra bit of configuration in order to be operational. This is now done
931 by default when running a QEMU domain with q35 machine type. This is not a
932 change in the guest ABI, but it is a guest visible behavior change since the
933 watchdog that did not fire before will now fire once used. To switch to the
934 previous behavior the watchdog action must be set to ``none``.
936 * QEMU: fix deleting memory snapshot when deleting external snapshots
938 When external snapshot deletion was introduced it did not remove memory
939 snapshot when it existed. In addition when external memory only snapshot
940 was created libvirt failed without producing any error.
942 * QEMU: properly report passt startup errors
944 Due to how the child passt process was started, the initial
945 support for passt (added in 9.0.0) would not see errors
946 encountered during startup, so libvirt would continue to setup and
947 start the guest; this led to a running guest with no network
950 (NB: On systems that use them, it is still necessary to disable
951 SELinux/AppArmor to start passt. This is a temporary limitation,
952 and use of the feature in production is strongly discouraged
953 until it has been lifted.)
955 * qemu: Fix error when attempting to change media in a CDROM drive
957 Due to a logic bug introduced in libvirt-9.0 attempts to change media in a
958 CDROM would previously fail with an error stating that the tray isn't open.
960 * qemu: Properly handle block job transitions
962 Starting with libvirt-9.0 the block job state machine improperly handled
963 some job transitions, which resulted into some block jobs not being
964 properly terminated. This could cause problems such as errors when
965 detaching a disk after snapshot.
967 * virsh: Make domif-setlink work more than once
969 There was a bug introduced in the previous release which made ``virsh
970 domif-setlink`` work exactly once over given domain. The bug was fixed and
971 now the command can be run multiple times.
973 * qemu: Make domain startup fail if NIC already exists
975 When starting a domain with an ``<interface/>`` that's supposed to be
976 managed by libvirt (``managed='yes'``) but corresponding TAP device already
977 exists, report an error and make the startup process fail.
979 * qemu: Deal with nested mounts when umount()-ing /dev
981 When setting up private ``/dev`` for a domain (also known as ``namespaces``
982 in ``qemu.conf``), libvirt preserves mount points nested under ``/dev``
983 (e.g. ``/dev/shm``, ``/dev/pts`` and so on). But there was a bug which
984 resulted in inability to construct the namespace when there were two or
985 more filesystems mounted on the same path. This is common scenario with
986 containers and thus the bug was fixed.
988 * remote: Pass ``mode`` and ``socket`` URI parameters to virt-ssh-helper
990 When connecting to a remote host using SSH transport, ``?mode=`` and
991 ``?socket=`` URI parameters were ignored. This prevented users from
992 connecting to a monolithic daemon running on a remote host.
994 * qemu: Various ``swtpm`` related fixes
996 There are more cleanups and small bug fixes with regards to emulated
997 ``<tpm/>``. For instance with migration when the ``swtpm`` state is on a
998 shared volume, or seclabel setting/restoring.
1006 * QEMU: implement external snapshot deletion
1008 External snapshot deletion is now possible using the existing API
1009 ``virDomainSnapshotDelete()``. Flags that allow deleting children
1010 or children only are not supported.
1012 * QEMU: support passt (https://passt.top)
1014 passt can be used to connect an emulated network device to the
1015 host's network without requiring libvirt to have any sort of
1016 elevated privileges. This is configured with::
1018 <interface type='user'>
1019 <backend type='passt'>
1022 * QEMU: add external backend for swtpm
1024 Connecting the VM to a swtpm daemon started outside of libvirt
1027 * QEMU: Support for passing FDs instead of opening files for `<disk>`
1029 A new API `virDomainFDAssociate` gives the users the option to pass FDs
1030 to libvirt and then use them when starting a VM. Currently the FDs can
1031 be used instead of directly opening files as `<disk>` backend.
1035 * qemu: Prefer PNG for domain screenshots
1037 With sufficiently new QEMU (v7.1.0) screenshots change format from PPM to PNG.
1039 * tools: Fix install_mode for some scripts
1041 Scripts from the following list were installed with group write bit set:
1042 virt-xml-validate, virt-pki-validate, virt-sanlock-cleanup,
1043 libvirt-guests.sh. This was changed so that only the owner is able to write
1046 * qemu: Allow multiple nodes for preferred policy
1048 Due to restrictions of old kernels and libnuma APIs, the preferred NUMA
1049 policy accepted just a single host NUMA node. With recent enough kernel
1050 (v5.15.0) and libnuma (v2.0.15) it's possible to set multiple nodes.
1052 * secret: Inhibit shutdown of daemon for ephemeral secrets
1054 When an ephemeral secret is defined then automatic shutdown of virtsecretd
1055 is inhibited. This is to avoid ephemeral secrets disappearing shortly
1058 * qemu: Report Hyper-V Enlightenments in domcapabilities
1060 The supported Hyper-V Enlightenments are now reported in domain
1065 * Fix NULL-pointer dereference `virXMLPropStringRequired`
1067 Fix a bug where when parsing a XML property which is required to be present
1068 by using `virXMLPropStringRequired` the parser will crash instead of
1071 * qemu: Init ext devices paths on reconnect
1073 Paths for external devices are not stored in the status XML. Therefore,
1074 when the daemon restarted and was reconnecting to a running domain, these
1075 paths were left blank which led to the daemon crash.
1077 * qemu: Validate arguments passed to `virConnectGetDomainCapabilities`
1079 There was a code path in which insufficient validation of input arguments
1080 of `virConnectGetDomainCapabilities` API was possible which led to the
1081 daemon crash. This path is now fixed.
1084 v8.10.0 (2022-12-01)
1085 ====================
1089 * Tool for validating SEV firmware boot measurement of QEMU VMs
1091 The ``virt-qemu-sev-validate`` program will compare a reported SEV/SEV-ES
1092 domain launch measurement, to a computed launch measurement. This
1093 determines whether the domain has been tampered with during launch.
1095 * Support for SGX EPC (enclave page cache)
1097 Users can add a ``<memory model='sgx-epc'>`` device to launch a VM with
1098 ``Intel Software Guard Extensions``.
1100 * Support migration of vTPM state of QEMU vms on shared storage
1102 Pass ``--migration`` option if appropriate in order for ``swtpm`` to
1103 properly migrate on shared storage.
1107 * Mark close callback (un-)register API as high priority
1109 High priority APIs use a separate thread pool thus can help in eliminating
1110 problems with stuck VMs. Marking the close callback API as high priority
1111 allows ``virsh`` to properly connect to the daemon in case the normal
1112 priority workers are stuck allowing other high priority API usage.
1114 * Updated x86 CPU features
1116 The following features for the x86 platform were added:
1117 ``v-vmsave-vmload``, ``vgif``, ``avx512-vp2intersect``, ``avx512-fp16``,
1118 ``serialize``, ``tsx-ldtrk``, ``arch-lbr``, ``xfd``, ``intel-pt-lip``,
1119 ``avic``, ``sgx``, ``sgxlc``, ``sgx-exinfo``, ``sgx1``, ``sgx2``,
1120 ``sgx-debug``, ``sgx-mode64``, ``sgx-provisionkey``, ``sgx-tokenkey``,
1121 ``sgx-kss``, ``bus-lock-detect``, ``pks``, ``amx``.
1123 * Add support for ``hv-avic`` Hyper-V enlightenment
1125 ``qemu-6.2`` introduced support for the ``hv-avic`` enlightenment which
1126 allows to use Hyper-V SynIC with hardware APICv/AVIC enabled.
1128 * qemu: Run memory preallocation with numa-pinned threads
1130 Run the thread allocating memory in the proper NUMA node to reduce overhead.
1132 * RPM packaging changes
1134 - add optional dependency of ``libvirt-daemon`` on ``libvirt-client``
1136 The ``libvirt-guests.`` tool requires the ``virsh`` client to work
1137 properly, but we don't want to require the installation of the daemon
1138 if the tool is not used.
1140 - relax required ``python3-libvirt`` version for ``libvirt-client-qemu``
1142 The ``virt-qemu-qmp-proxy`` tool requires python but doesn't strictly
1143 need the newest version. Remove the strict versioning requirement in
1144 order to prevent cyclic dependency when building.
1148 * Skip initialization of ``cache`` capabilities if host doesn't support them
1150 Hypervisor drivers would fail to initialize on ``aarch64`` hosts with
1153 virStateInitialize:657 : Initialisation of cloud-hypervisor state driver failed: no error
1155 which prevented the startup of the daemon.
1157 * Allow incoming connections to guests on routed networks w/firewalld
1159 A change in handling of implicit rules in ``firewalld 1.0.0`` broke
1160 incoming connections to VMs when using ``routed`` network. This is fixed
1161 by adding a new ``libvirt-routed`` zone configured to once again allow
1162 incoming sessions to guests on routed networks.
1164 * Fix infinite loop in nodedev driver
1166 Certain udev entries might be of a size that makes libudev emit EINVAL
1167 which caused a busy loop burning CPU. Fix it by ignoring the return code.
1175 * Add ``virt-qemu-qmp-proxy`` for emulating a QMP socket for libvirt managed VMs
1177 ``virt-qemu-qmp-proxy`` tool provides a way to expose an emulated QMP server
1178 socket for a VM managed by libvirt. This allows existing QMP-only clients
1179 to work with libvirt managed VMs.
1181 **Note:** libvirt is not interpreting the communication between the tool
1182 using the proxy and qemu itself, so any state-changing commands may
1183 desynchronize libvirt. Use at your own risk.
1185 * qemu: Core Scheduling support
1187 To avoid side channel attacks, the Linux kernel allows creating groups of
1188 processes that trust each other and thus can be scheduled to run on
1189 hyperthreads of a CPU core at the same time. This is now implemented for
1190 QEMU domains too (see ``sched_core`` knob in qemu.conf), although not
1191 enabled by default, just yet.
1195 * qemu: Add hypervisor-specific statistics to ``virConnectGetAllDomainStats``
1197 The new stats group ``VIR_DOMAIN_STATS_VM`` of
1198 ``virConnectGetAllDomainStats``, also exposed as ``virsh domstats --vm``,
1199 returns hypervisor-specific stats fields for given VM.
1201 * Add ``vendor`` attribute for CPU models in domain capabilities
1203 Users can now see the vendor of each CPU model in domain capabilities and
1204 use it, e.g., for filtering usable CPU models based on host CPU vendor.
1206 * virsh: Add ``--model`` option for ``hypervisor-cpu-baseline``
1208 This is a shortcut for calling ``hypervisor-cpu-baseline`` with a single
1209 CPU model and no additional features. It can be used for determining which
1210 features block a particular CPU model from being usable.
1212 * Improved documentation of CPU ``usable`` attribute in domain capabilities
1214 * Report ``channel`` and ``redirdev`` devices in domain capabilities
1216 The channel and redirect devices supported by the hypervisor are now
1217 reported in domain capabilities.
1219 * meson: Bump minimal required meson version
1221 Newer meson versions deprecate some functions used. These were replaced
1222 with their newer counterparts and the minimal required mesion version was
1225 * qemu: Add flags to keep or remove TPM state for ``virDomainUndefineFlags``
1227 ``VIR_DOMAIN_UNDEFINE_TPM`` and ``VIR_DOMAIN_UNDEFINE_KEEP_TPM`` specify
1228 accordingly to delete or keep a TPM's persistent state directory structure
1229 and files when undefining a domain. In virsh the flags are exposed as
1230 ``--tpm`` and ``--keep-tpm`` for the sub-command ``undefine``.
1234 * qemu: Disable all blocker features in CPU baseline
1236 Three years ago QEMU renamed some CPU features (mostly those containing
1237 an underscore). When such renamed feature was reported by QEMU as blocking
1238 usability of a CPU model, we would fail to explicitly disable it when
1239 creating a baseline CPU definition using this model. This bug did not have
1240 any functional impact when the default ``check='partial'`` attribute was
1241 used for guest CPU definition in domain XML, but it could have caused
1242 failures to start a domain with ``check='full'`` in some cases.
1244 * qemu: Do not crash after restart with active migration
1246 In 8.8.0 release libvirt daemon would crash after it was restarted during
1247 an active outgoing migration.
1249 * qemu: Refresh state after restore from a save image
1251 When a domain is restored from a saved image, libvirt now queries QEMU for
1252 those parts of runtime information that were not part of the save image.
1253 For instance: MAC address of a macvtap NICs, tray state of CD-ROMs,
1254 allocated size of virtio-mem, and others.
1260 * **Removed features**
1262 * storage: Remove 'sheepdog' storage driver backend
1264 The 'sheepdog' project is no longer maintained and upstream bug reports
1265 are unaddressed. Libvirt thus removed the support for the sheepdog storage
1266 driver backend, following qemu's removal of sheepdog support in qemu-6.1.
1270 * qemu: Implement VIR_DOMAIN_STATS_CPU_TOTAL for qemu:///session
1272 Users can now query VIR_DOMAIN_STATS_CPU_TOTAL (also known as cpu.time)
1273 statistics for session domains.
1277 * qemu: Fix non-shared storage migration setup
1279 This release fixes a bug in setup of a migration with non-shared storage
1280 ( ``virsh migrate --copy-storage-all``) which was broken by a refactor of
1281 the code in libvirt-8.7.
1283 * selinux: Don't ignore NVMe disks when setting image label
1285 Libvirt did not set any SELinux label on NVMe disks and relied only on the
1286 default SELinux policy. This turned out to cause problem when using
1287 namespace or altered policy and thus is fixed now.
1289 * qemu: Fix a deadlock when setting up namespace
1291 When starting a domain, libvirt creates a mount namespace and manages
1292 private /dev with only a handful nodes exposed. But when creating those a
1293 deadlock inside glib might have occurred. The code was changed so that
1294 libvirt does not tickle the glib bug.
1296 * qemu: Don't build memory paths on daemon restart
1298 When the daemon is restarted it tried to create domain private paths for
1299 each mounted hugetlbfs. When this failed, the corresponding domain was
1300 killed. This operation is now performed during domain startup and memory
1301 hotplug and no longer leads to sudden kill of the domain.
1307 * **Removed features**
1309 * qemu: Remove support for QEMU < 4.2
1311 In accordance with our platform support policy, the oldest supported QEMU
1312 version is now bumped from 3.1 to 4.2.
1316 * qemu: Add support for specifying vCPU physical address size in bits
1318 Users can now specify the number of vCPU physical address bits with
1319 the `<maxphysaddr>` subelement of the `<cpu>` element.
1323 * esx: Domain XMLs can now be dumped for VMs with two new interface types
1325 One is when the interface is not connected anywhere `type='null'` and one
1326 when it is connected to VMWare Distributed Switch `type='vds'`.
1330 * qemu: increase memlock limit for a domain with multiple vfio/vdpa devices
1332 When multiple vfio or vdpa devices are assigned to a domain, the locked
1333 memory limit could be too low to map memory for all devices. The memlock
1334 limit has been increased to be proportional to the number of vdpa/vfio
1343 * conf: Improved firmware autoselection
1345 The firmware autoselection feature now behaves more intuitively, reports
1346 better error messages on failure and comes with high-level documentation.
1354 * qemu: Introduce support for network backed NVRAM
1356 Users can now use remote store NVRAM image by specifying newly introduced
1357 attribute `type='network'` with `<nvram>` element.
1359 * qemu: Add support for post-copy migration recovery
1361 A new ``VIR_MIGRATE_POSTCOPY_RESUME`` flag (``virsh migrate --postcopy-resume``)
1362 was introduced for recovering from a failed post-copy migration.
1364 * qemu: Add support for zero-copy migration
1366 With QEMU 7.1.0, libvirt can enable zerocopy for parallel migration. This
1367 is implemented by adding a new ``VIR_MIGRATE_ZEROCOPY`` flag(``virsh migrate
1370 * Introduce thread_pool_min and thread_pool_max attributes to IOThread
1372 New attributes ``thread_pool_min`` and ``thread_pool_max`` were introduced
1373 to ``<iothread/>`` as well as new ``<defaultiothread/>`` element with the
1374 same attributes. This way it's possible to instruct QEMU to spawn enough
1375 worker threads for an IOThread upfront, resulting in predictable time
1376 needed to process an I/O request.
1380 * Define a TFTP server without a DHCP server in network configuration
1382 It's now possible to define a network with no DHCP server but with a TFTP
1383 server. This may be useful when DHCP service is provided by other entity on
1384 the network than libvirt spawned dnsmasq.
1388 * qemu: Restore label to temp file in qemuDomainScreenshot()
1390 When virDomainScreenshot() is called, libvirt instructs QEMU to save the
1391 screenshot into a temporary file. This file needs to be labelled correctly,
1392 so that QEMU can access it. And since the file is temporary (it's deleted
1393 after the screenshot was taken) the corresponding label restore was
1394 missing. This proven to be problematic for profile based models, like
1395 AppArmor, where the temporary files were added into the profile but never
1396 removed, which resulted in longer profile recalculation times.
1398 * qemuBuildInterfaceConnect: Initialize @tapfd array
1400 Due to an uninitialized array, unsuccessful attempt to start a guest with
1401 an ``<interface/>`` might have resulted in closing of a random FD and thus
1402 sudden disconnect of a client or other random failures.
1404 * qemu: Fix hotplug of network interfaces
1406 A logic bug introduced in a recent refactor was fixed. The bug caused a
1407 problem when hot-adding a network interface, which failed with the
1410 error: internal error: unable to execute QEMU command 'netdev_add': File descriptor named '(null)' has not been found
1412 * Fix ``startupPolicy`` validation for ``block`` disks
1414 Setting of ``startupPolicy`` for a block disk would result in an error due
1415 to a logic bug in a recent refactor.
1417 * qemu: Fix crash when overriding device properties via ``<qemu:override>`` element
1419 Adding an override for a device property would result in a crash of the qemu
1428 * qemu: D-Bus display
1430 Libvirt is now able to setup a D-Bus display export, either with a private
1431 bus or in p2p mode. This display is available in QEMU 7.0.0.
1433 * qemu: ppc64 Power10 processor support
1435 Support for the recently released IBM Power10 processor was added.
1437 * qemu: Introduce ``absolute`` clock offset
1439 The ``absolute`` clock offset type allows to set the guest clock to an
1440 arbitrary epoch timestamp at each start. This is useful if some VM needs
1441 to be kept set to an arbitrary time for e.g. testing or working around
1444 * qemu: add qemu-vdagent channel
1446 This paravirtualized qemu vdagent channel can enable copy and paste between
1447 a guest and a VNC client. It is available in QEMU 6.1.0.
1449 * api: Add new APIs ``virDomainSaveParams`` and ``virDomainRestoreParams``
1451 * ``virDomainSaveParams``: An alternative domain saving API, extends
1452 ``virDomainSaveFlags`` by adding parameters.
1453 * ``virDomainRestoreParams``: An alternative domain restoring API, extends
1454 ``virDomainRestoreFlags`` by adding parameters.
1458 * Improve heuristics for computing baseline CPU models
1460 Both ``virConnectBaselineHypervisorCPU`` and ``virConnectBaselineCPU`` were
1461 in some cases computing the result using a CPU model which was newer than
1462 some of the input models. For example, ``Cascadelake-Server`` was used as a
1463 baseline for ``Skylake-Server-IBRS`` and ``Cascadelake-Server``. The CPU
1464 model selection heuristics was improved to choose a more appropriate model.
1470 * **Removed features**
1472 * qemu: Remove support for QEMU < 3.1
1474 In accordance with our platform support policy, the oldest supported QEMU
1475 version is now bumped from 2.11 to 3.1.
1479 * qemu: Introduce support for virtio-iommu
1481 This IOMMU device can be used with both Q35 and ARM virt guests.
1483 * qemu: Introduce attributes rss and rss_hash_report for net interface
1485 They can enable in-qemu/ebpf RSS and in-qemu RSS hash report for virtio NIC.
1486 Require QEMU >= 5.1.
1494 * qemu: Introduce ``manual`` disk snapshot mode
1496 This new mode allows users to synchronize libvirt snapshots with snapshots
1497 which need to be done outside of libvirt e.g. when 'vhost-user-blk' is used
1500 * Introduce memory allocation threads
1502 When starting a QEMU guest, libvirt can now instruct QEMU to allocate
1503 guest's memory in parallel. This may be handy when guest has large amounts
1508 * qemu: ``VIR_MIGRATE_PARAM_TLS_DESTINATION`` now works with non-shared storage migration
1510 The setting now also applies to the NBD connections for non-shared storage
1511 migration allowing migration to proceed even when the user expects certificate
1514 * qemu: Allow overrides of device properties via the qemu namespace
1516 Users wishing to override or modify properties of devices configured by
1517 libvirt can use the ``<qemu:deviceOverride>`` QEMU namespace element to
1518 specify the overrides instead of relying on the argv passthrough of the
1519 ``-set`` qemu commandline option which no longer works with new qemu.
1521 * qemu: Allow passing file descriptors to ``virsh qemu-monitor-command``
1523 Passing FDs allows users wanting to experiment with qemu driven by libvirt
1524 use commands like ``add-fd`` properly.
1526 * libxl: Turn on user aliases
1528 Users can now use so called user aliases for XEN domains.
1530 * Implement support for FUSE3
1532 The LXC driver uses fuse to overwrite some lines in ``/proc/meminfo``
1533 inside containers so that they see correct amount of memory given to them.
1534 The code was changed so that both ``fuse`` and ``fuse3`` are supported.
1536 * Improve domain save/restore throughput
1538 Code that's handling save or restore of QEMU domains was changed resulting
1539 in better performance of I/O and thus shortening time needed for the operation.
1543 * Both build and tests should now pass on Alpine Linux or any other
1544 distribution with musl libc.
1546 * virsh: Fix integer overflow in allocpages
1548 On hosts which support hugepages larger than 1GiB ``virsh allocpages``
1549 failed to accept them because of an integer overflow. This is now fixed.
1551 * qemu: Fix segmentation fault in virDomainUndefineFlags
1553 When a domain without any ``<loader/>`` was being undefined, libvirt has
1554 crashed. This is now fixed.
1556 * lxc: Fix unaligned reads of /proc/meminfo within a container
1558 When /proc/meminfo was read in chunks smaller than the entire file, libvirt
1559 would produce mangled output. While porting the code to FUSE3 this area was
1560 reworked and the file can now be read with any granularity.
1562 * qemu: Be less aggressive around cgroup_device_acl
1564 A basic set of devices common to every domain can be set in ``qemu.conf``
1565 via cgroup_device_acl knob. Devices from this set are allowed in CGroup and
1566 created in domain private namespace for every domain. However, upon device
1567 hotunplug it may have had happened that libvirt mistakenly denied a device
1568 from this set and/or removed it from the namespace. For instance,
1569 /dev/urandom was removed and denied in CGroup on RNG hotunplug.
1571 * nodedev: trigger mdev device definition update on udev add and remove
1573 When nodedev objects are added and removed mdev device definitions are
1574 updated to report correct associated parent.
1582 * qemu: Add hvf domain type for Hypervisor.framework
1584 It works on Intel machines as well as recent machines powered by Apple
1585 Silicon. QEMU 6.2.0 is needed for Apple Silicon support.
1587 * qemu: Support mode option for dirtyrate calculation
1589 Introduce ``virDomainDirtyRateCalcFlags`` as parameter of
1590 ``virDomainStartDirtyRateCalc``, which is used to specify the mode of
1591 dirty page rate calculation.
1593 Add ``--mode`` option to ``virsh domdirtyrate-calc``, which can be
1594 either of the following 3 options:
1595 ``page-sampling, dirty-bitmap, dirty-ring``.
1597 Add ``calc_mode`` field for dirtyrate statistics returned by
1598 ``virsh domstats --dirtyrate``, also add ``vCPU dirtyrate`` if
1599 ``dirty-ring`` mode was used in last measurement.
1603 * packaging: sysconfig files no longer installed
1605 libvirt used to provide defaults in various /etc/sysconfig/ files, such
1606 as /etc/sysconfig/libvirtd. Since these files are owned by the admin, this
1607 made it difficult to change built-in defaults in case such file was
1608 modified by the admin. The built-in defaults are now part of the provided
1609 systemd unit files, such as libvirtd.service. These unit files continue
1610 to parse sysconfig files, in case they are created by the admin and filled
1611 with the desired key=value pairs.
1613 * virnetdev: Ignore EPERM on implicit clearing of VF VLAN ID
1615 Libvirt will now ignore EPERM errors on attempts to implicitly clear a
1616 VLAN ID (when a VLAN is not explicitly provided via an interface XML
1617 using a 0 or a non-zero value) as SmartNIC DPUs do not expose VLAN
1618 programming capabilities to the hypervisor host. This allows Libvirt
1619 clients to avoid specifying a VLAN and expect VF configuration to work
1620 since Libvirt tries to clear a VLAN in the same operation
1621 as setting a MAC address for VIR_DOMAIN_NET_TYPE_HOSTDEV devices which
1622 is now split into two distinct operations. EPERM errors received while
1623 trying to program a non-zero VLAN ID or explicitly program a VLAN ID 0
1624 will still cause errors as before so there is no change in behavior
1629 * Remove unix sockets from filesystem when disabling a '.socket' systemd unit
1631 The presence of the socket files is used by our remote driver to determine
1632 which service to access. Since neither systemd nor the daemons clean up the
1633 socket file clients were running into problems when a modular deployment was
1634 switched to monolithic ``libvirtd``.
1636 * qemu: Fixes of fd passing during hotplug and hotunplug of chardevs
1638 FDs used as chardev backing are now properly removed when hot-unplugging
1639 a chardev from qemu and hotplugged chardevs now properly use ``virtlogd``
1640 to handle the input and output from qemu.
1642 * RPM: Run pre/post-install steps on ``daemon-driver-storage-core``
1644 Previously the pre/post-install code was part of the meta-package which
1645 installed all storage driver sub-packages thus a minimalistic install
1646 of the storage driver didn't behave correctly.
1654 * libxl: Fix potential deadlock and crash (CVE-2021-4147)
1656 A rogue guest could continuously reboot itself and cause libvirtd on the
1657 host to deadlock or crash, resulting in a denial of service condition.
1659 * **Removed features**
1661 * qemu: Explicitly forbid live changing nodeset for strict numatune
1663 For ``strict`` mode of <numatune/> it can't be guaranteed that memory is
1664 moved completely onto new set of nodes (e.g. QEMU might have locked pieces
1665 of its memory) thus breaking the strict promise. If live migration of QEMU
1666 memory between NUMA nodes is desired, users are advised to use
1667 ``restrictive`` mode instead.
1671 * qemu: Synchronous write mode for disk copy operations
1673 The ``blockdev-mirror`` block job supports a mode where writes from the VM
1674 are synchronously propagated to the destination of the copy. This ensures
1675 that the job will converge under heavy I/O.
1677 Implement the mode for the copy blockjob as
1678 ``VIR_DOMAIN_BLOCK_COPY_SYNCHRONOUS_WRITES`` flag exposed via
1679 ``virsh blockcopy --synchronous-writes`` and for non-shared storage migration
1680 as ``VIR_MIGRATE_NON_SHARED_SYNCHRONOUS_WRITES`` exposed via
1681 ``virsh migrate --copy-storage-synchronous-writes``.
1683 * Introduce TCG domain features
1685 Libvirt is now able to set the size of translation block cache size
1686 (tb-size) for TCG domains.
1688 * qemu: Add new API to inject a launch secret in a domain
1690 New API ``virDomainSetLaunchSecurityState()`` and virsh command
1691 ``domsetlaunchsecstate`` are added to support injecting a launch secret
1692 in a domain's memory.
1696 * libxl: Implement the virDomainGetMessages API
1698 * qemu: Preserve qcow2 sub-cluster allocation state after external snapshots and block-copy
1700 The new image which is installed as an overlay on top of the current chain
1701 when taking an external snapshot, or the target of a block copy operation
1702 now enables sub-cluster allocation (``extended_l2``) if the original
1703 image has the option enabled.
1707 * qemu: Fix device hot-unplug with ``libvirt-7.9`` or ``libvirt-7.10`` used with ``qemu-6.2``
1709 An internal change to the configuration format used by the above libvirt
1710 versions triggers a bug in ``qemu-6.2`` where qemu no longer emits the
1711 event notifying that the device was unplugged successfully and thus libvirt
1712 never removes the device from the definition.
1714 This impacts only devices which were present at startup of the VM, hotplugged
1715 devices behave correctly.
1717 This is fixed in ``libvirt-8.0`` by reverting to the old configuration
1718 approach until qemu is fixed.
1720 As a workaround for ``libvirt-7.9`` and ``libvirt-7.10`` the old configuration
1721 approach can be forced by:
1723 Option 1, global ``qemu.conf``::
1725 capability_filters = [ "device.json" ]
1727 Option 2, per VM XML override::
1729 <domain type='kvm' xmlns:qemu='http://libvirt.org/schemas/domain/qemu/1.0'>
1734 <qemu:del capability='device.json'/>
1735 </qemu:capabilities>
1738 * Fix sparse streams with split daemon
1740 In split daemon scenario, a client connected to a hypervisor driver and
1741 using sparse streams (e.g. ``virsh vol-download --sparse``) would make the
1742 hypervisor daemon enter an infinite loop without any data transfer. This is
1745 * Build no longer requires RPC library
1747 Code and its cross dependencies were fixed so that build without remote
1748 driver and thus an RPC library (like ``tirpc``) fails no more.
1750 * virnetdevopenvswitch: Fix 'burst' value passed to ovs-vsctl
1752 When a ``<bandwidth/>`` was defined for a TAP device that's plugged into an
1753 OvS bridge values passed to the OvS were incorrectly recalculated resulting
1754 in slightly different limits being applied.
1757 v7.10.0 (2021-12-01)
1758 ====================
1762 * Added virt-pki-query-dn binary
1764 This binary helps users figure out the format of Distinguished Name
1765 from a certificate file the way that libvirt expects it in
1766 tls_allowed_dn_list option of libvirtd.conf configuration file
1770 * qemu: Report guest interface information in ``virDomainGetGuestInfo``
1772 Libvirt is now able to report interface information from the guest's
1773 perspective (using guest agent).
1775 * qemu: detect guest side errors during device removal
1777 Libvirt is now able to detect guest side errors during device removal by
1778 using the DEVICE_UNPLUG_GUEST_ERROR event, available in QEMU 6.2.0.
1780 * Minimum SSF setting
1782 The libvirtd.conf option tcp_min_ssf can be used to override the minimum
1783 permitted security strength factor for non-TLS remote connections.
1784 The current hardcoded minimum is 56 (single-DES) and will be raised to 112
1785 in the future. Setting a minimum lower than 112 is not supported.
1787 * qemu: Report stats also for block copy destination and backup job scratch
1790 The statistics are available via the bulk domain stats API.
1794 * qemu: Don't format 'ramfb' attribute when disabled
1796 Fix a regression caused by the conversion to JSON -device arguments where
1797 'ramfb' would be put on the commandline of 'vfio-pci' which doesn't have it,
1798 thus breaking VMs with a mediated host device.
1800 * qemu: Fix block copy and backup to encrypted storage
1802 An oversight in last release lead to a spurious error being reported when
1803 encrypted storage was requested for disk images which are not directly
1804 backing the disk, which is now fixed.
1812 * Introduce virtio-mem ``<memory/>`` model
1814 New virtio-mem model is introduced for ``<memory/>`` device which is a
1815 paravirtualized mechanism of adding/removing memory to/from a VM. Use
1816 ``virDomainUpdateDeviceFlags()`` API to adjust amount of memory or ``virsh
1817 update-memory-device`` for convenience.
1819 * qemu: support disabling hotplug of devices on the pci-root controller
1821 the <target hotplug='on|off'/> option is now supported for the
1822 pci-root controller on i440fx-based (x86 "pc") machinetypes. This
1823 can be used to disable hotplug/unplug of devices from this
1824 controller. The default behavior is unchanged (hotplug is
1827 * Support hotplug and hotunplug for virtiofs
1829 Filesystems backed by virtiofsd can now be hotplugged and hotunplugged.
1831 * virpcivpd: Add a PCI VPD parser
1833 A parser for the standard PCI/PCIe VPD ("I.3. VPD Definitions" in PCI 2.2+
1834 and an equivalent definition in "6.28.1 VPD Format" PCIe 4.0) was added
1835 along with relevant types to represent PCI VPD in memory. This
1836 functionality got added for Linux only at this point (kernels above
1837 v2.6.26 have support for exposing VPD via sysfs).
1839 * virpci: Add PCI VPD-related helper functions to virpci
1841 In order to utilize the PCI VPD parser, a couple of helper functions got
1842 introduced to check for the presence of a VPD file in the sysfs tree and
1843 to invoke the PCI VPD parser to get a list of resources representing PCI
1844 VPD contents in memory.
1846 * nodedev: Add PCI VPD capability support
1848 Support for serializing and deserializing PCI VPD data structures is added
1849 following the addition of the PCI VPD parser. A new PCI device capability
1850 called "vpd" is introduced holding string resources and keyword resources
1853 * qemu: Support page_per_vq for driver element
1855 This optional virtio attribute ``page_per_vq`` controls the layout of the
1856 notification capabilities exposed to the guest. It is recommended for the
1859 * qemu: Support librbd encryption
1861 Add an encryption engine ``librbd``. It will provides the image-level
1862 encryption of librbd. It requires QEMU >= 6.1.0 and librbd >= 16.1.0.
1866 * Use of JSON syntax with ``-device`` with upcoming QEMU-6.2
1868 Libvirt started using JSON directly with the ``-device`` commandline
1869 parameter as it's considered the preferred stable syntax for further QEMU
1870 releases. If any problems with the conversion are encountered please
1871 report them as soon as possible.
1875 * qemu: Fix problems on ``virsh domstats`` with qemu <5.2.0
1877 Libvirt v7.2.0 and later called query-dirty-rate, which was introduced in
1878 qemu-5.2.0, regardless of qemu version and failed in qemu-5.1.0. This
1879 release fixes the bug.
1881 * Don't enter endless loop when unable to accept new clients
1883 If libvirtd (or any other daemon) hit the ulimit for maximum number of open
1884 files but there are still client connections pending then libvirtd (or
1885 corresponding split daemon) would enter an endless loop from which it would
1886 never recover. This behaviour is now fixed.
1888 * qemu: Run secondary driver hooks in split daemon mode
1890 Because of a bug in implementation it may happen that hooks from secondary
1891 drivers were not called in all cases, for instance a network hook wasn't
1892 called upon removal of interface after domain shut off itself. With this
1893 release the bug is fixed.
1901 * nodedev: Add ability to automatically start mediated devices
1903 The autostart status of a persistent mediated devices can be managed with
1904 the new APIs ``virNodeDeviceSetAutostart()`` and
1905 ``virNodeDeviceGetAutostart()``. The corresponding virsh command is
1906 ``nodedev-autostart``. In addition, two new APIs were added to get
1907 additional information about node devices: ``virNodeDeviceIsPersistent()``
1908 checks whether the device is persistently defined, and
1909 ``virNodeDeviceIsActive()`` checks whether the node device is currently
1910 active. This information can also be retrieved with the new virsh command
1913 * qemu: Add attribute ``queue_size`` for virtio-blk devices
1917 * api: Add XML validation for creating of: networkport, nwfilter-binding,
1920 * Add flag ``VIR_NETWORK_PORT_CREATE_VALIDATE`` to validate network port
1921 input xml of network-port creating.
1922 * Add flag ``VIR_NETWORK_CREATE_VALIDATE`` to validate network input xml of
1924 * Add flag ``VIR_NWFILTER_BINDING_CREATE_VALIDATE`` to validate
1925 nwfilter-binding input xml of nwfilter-binding creating.
1933 * Add support for Fibre Channel VMID
1935 New VM element ``<fibrechannel appid=''/>`` was added to allow users to set
1936 their ``appid`` for each VM which will be used by kernel to create Fibre
1937 Channel VMID. This allows various QoS levels, access control or collecting
1938 telemetry data per VM.
1942 * virsh: Allow XML validation for define of: storage pool, network, secret,
1945 * Add flag ``VIR_STORAGE_POOL_DEFINE_VALIDATE`` to validate storage pool
1946 input xml. For virsh, users can use it as ``virsh pool-define --validate``.
1947 * Add flag ``VIR_NETWORK_DEFINE_VALIDATE`` to validate network input xml. For
1948 virsh, users can use it as ``net-define --validate``.
1949 * Add flag ``VIR_SECRET_DEFINE_VALIDATE`` to validate secret input xml. For
1950 virsh, users can use it as ``secret-define --validate``.
1951 * Add flag ``VIR_NWFILTER_DEFINE_VALIDATE`` to validate nwfilter input xml.
1952 For virsh, users can use it as ``nwfilter-define --validate``.
1953 * Add flag ``VIR_INTERFACE_DEFINE_VALIDATE`` to validate interface input xml.
1954 For virsh, users can use it as ``iface-define --validate``.
1956 * Add SecurityManager APIs for labeling network devices
1958 New ``virSecurityManagerSetNetdevLabel`` and ``virSecurityManagerSetNetdevLabel``
1959 APIs are introduced and implemented in the Apparmor security driver.
1960 The qemu driver uses the APIs to label vhostuser ports on hotplug and
1961 restore labeling on unplug.
1963 * vmx: Parse vm.genid and support super wide SCSI bus
1965 The genid attribute is now reported for VMX guests. Libvirt can now
1966 properly process super wide SCSI bus (64 units).
1968 * qemu: Lifecycle action (``on_poweroff``/``on_reboot``) handling improvements
1970 The handling of lifecycle actions was fixed and improved in multiple ways:
1972 - ``restart-rename`` action was forbidden
1974 The action was never properly implemented in the qemu driver and didn't
1975 actually result in a restart of the VM but rather termination. The qemu
1976 driver now rejects such configurations.
1978 - ``preserve`` action was forbidden
1980 Similarly to the previous case this never worked as the intended semantics
1981 of the actions dictate. It's better to not allow it at all until there's a
1982 proper implementation
1984 - ``reboot`` action of ``on_poweroff`` now actually works
1986 The guest OS is now rebooted instead of terminating the VM when the
1987 ``reboot`` action is used and the guest OS powers down. Note that it's
1988 incompatible with ``on_reboot`` set to ``destroy``.
1990 - Changes in action action of ``on_reboot`` are now updated with qemu
1992 Libvirtd can now properly update the ``on_reboot`` action in qemu which
1993 allows proper handling when changing between ``reboot`` and ``destroy``
1994 actions. In addition, switching from ``reboot`` to ``destroy`` was
1995 forbidden for older qemus which don't support the update API as the guest
1996 could still reboot and execute some instructions until it was terminated.
1998 * virsh: Support vhostuser in attach-interface
2002 * qemu: Open chardev logfile on behalf of QEMU
2004 Guests with a logfile configured for their chardevs are now able to start
2005 even when no virtlogd is configured.
2007 * virhostmem: Handle numactl-less build in hugepages allocation/reporting
2009 Some architectures don't have notion of NUMA (e.g. s390x) but do support
2010 hugepages. Libvirt silently ignored requests to allocate/report hugepage
2011 pool when built without numactl. This is now fixed and the pool can be
2012 allocated/reported on properly.
2014 * qemu: Record proper ``backing`` format for overlays of qcow2+luks images
2016 Libvirt would record ``luks`` instead of ``qcow2`` into the metadata. In
2017 practice this is a problem only when inspecting images manually via
2018 ``qemu-img`` as with libvirt users must use full specification of the
2019 backing chain in the domain XML which supersedes information recorded in
2028 * storage: Unlock pool objects on ACL check failures in ``storagePoolLookupByTargetPath`` (CVE-2021-3667)
2030 A logic bug in ``storagePoolLookupByTargetPath`` where the storage pool
2031 object was left locked after a failure of the ACL check could potentially
2032 deprive legitimate users access to a storage pool object by users who don't
2037 * qemu: Incremental backup support via ``virDomainBackupBegin``
2039 libvirt-7.6 along with the unreleased qemu-6.1 will fully support the change
2040 block tracking features (block-dirty-bitmaps) to be able to do incremental
2041 backups and management of the checkpoint states via the appropriate APIs.
2043 * qemu: Add support for launch security type s390-pv
2045 Specifying s390-pv as launch security type in an s390 domain prepares for
2046 running the guest in protected virtualization secure mode, also known as
2047 IBM Secure Execution. This simplifies the definition and reduces the risk
2048 of an incorrect definition, e.g. by forgetting to specify ``iommu=on`` on
2051 * domstats: Add haltpolling time statistic interface
2053 Domstats now provide the data of cpu haltpolling time. This feature relies
2054 on statistics available after kernel version 5.8. This will allow the user
2055 to get more accurate CPU usage information if needed.
2059 * qemu: Fix migration with ``VIR_MIGRATE_NON_SHARED_INC``
2061 libvirt 7.3.0 introduced a bug where ``VIR_MIGRATE_NON_SHARED_INC`` would
2062 not actually migrate the contents of the disk due to broken logic and at
2063 the same time could trigger migration of storage when
2064 ``VIR_MIGRATE_TUNNELLED`` is requested. This release fixes the bug.
2066 * qemu: Don't emit ``VIR_DOMAIN_EVENT_ID_BLOCK_THRESHOLD`` twice when registered with index
2068 When registering the threshold event with the index notation (e.g.
2069 ``vda[3]``) libvirt would emit the event also for ``vda`` if the image is
2070 in the top layer. The intention was to emit two events only when the
2071 original registration was done without the index.
2073 * qemu: Pass discard requests for disks with ``copy_on_read='on'``
2075 When a disk using the ``copy_on_read='on'`` option is configured also with
2076 ``discard='unmap'`` the discard requests will now be passed to the
2077 underlying image freeing up the space.
2085 * svirt: fix MCS label generation (CVE-2021-3631)
2087 A flaw in the way MCS labels were generated could result in a VM's
2088 resource not being fully protected from access by another VM were
2089 it to be compromised. https://gitlab.com/libvirt/libvirt/-/issues/153
2091 * **Removed features**
2093 * xen: Remove support for Xen < 4.9
2095 In accordance with our platform support policy, the oldest supported Xen
2096 version is now bumped from 4.6 to 4.9.
2100 * docs: Document disk serial truncation status quo
2102 Disk ``<serial>`` is being truncated by QEMU before passed to the guest.
2103 Since it's impossible to fix it without running into further regressions
2104 the documentation was improved to document the intricacies.
2108 * qemu: Fixed validation of disk ``iothread`` configuration
2110 The validation of ``iothread`` config was previously moved to a place where
2111 it caused bogus errors when address wasn't allocated when hotplugging a
2112 disk. The check is now removed as it wasn't actually necessary at all.
2118 * **Removed features**
2120 * qemu: Remove support for QEMU < 2.11
2122 In accordance with our platform support policy, the oldest supported QEMU
2123 version is now bumped from 1.5 to 2.11.
2127 * qemu: Add support for hotplugging ``<transient/>`` disks
2129 The disk hotplug code in the qemu driver now can handle hotplug of disks
2130 with automatically added overlay.
2132 * qemu: Add support for sharing base image of ``<transient/>`` disks
2134 Users can use ``<transient shareBacking='yes'/>`` to tell the qemu driver to
2135 never open the base image in write mode thus multiple VMs can share the
2136 same image. Note that the disk will be hotplugged during startup.
2140 * Add win-dmp crashdump format
2142 New ``win-dmp`` format for ``virDomainCoreDumpWithFormat`` API and/or virsh
2143 ``dump --format`` was introduced.
2147 * Allow 0 offset in XML schema for ``<slice type='storage' offset='0' size='321'/>``
2149 Having a 0 offset so that the size of the image can be limited is a
2150 valid configuration so it was allowed in the XML schema.
2158 * xen: Support domains with more than 4TB
2160 The xen driver now supports domains with more than 4TB of memory with
2163 * qemu: add socket for virtiofs filesystems
2165 Libvirt now supports ``filesystem`` devices that connect to
2166 a ``virtiofsd`` daemon launched outside of libvirtd, via the
2167 ``socket`` attribute of the ``source`` element.
2169 * nodedev: Add ability to manage persistent mediated devices
2171 Persistent mediated devices can now be managed with libvirt.
2172 ``virNodeDeviceDefineXML()`` defines a new device,
2173 ``virNodeDeviceUndefine()`` removes an existing definition, and
2174 ``virNodeDeviceCreate()`` starts a device definition that is currently
2175 inactive. Corresponding virsh commands ``nodedev-define``,
2176 ``nodedev-undefine``, and ``nodedev-start`` were also added.
2177 ``nodedev-list`` only lists active devices by default. Inactive device
2178 definitions can be shown with the new ``--inactive`` and ``--all`` flags.
2180 * qemu: Allow use of qemu's ``-compat`` option
2182 Curious developers or testers now can enable certain ``-compat`` modes which
2183 allow to notice use of deprecated commands and options as qemu will use the
2184 selected method to notify the user. The new behaviour can be requested using
2185 either the ``deprecation_behavior`` option in ``qemu.conf`` for all VMs or
2186 using ``<qemu:deprecation behavior='...'/>`` in the VM XML.
2190 * virsh: Improve errors with ``virsh snapshot-create-as``
2192 The XML document constructed by virsh was forced through XML schema
2193 validation which yielded unintelligible error messages in cases such as
2194 when the path to the new image did not start with a slash. XML documents
2195 are no longer validated as the XML parser actually has better error
2196 messages which allow users to figure the problem out quickly.
2198 * qemu: Terminate backing store when doing a full-chain block pull
2200 When pulling everything into the overlay image the chain can be terminated
2201 since we know that it won't depend on any backing image and thus can prevent
2202 attempts to probe the backing chain.
2204 * qemu: Expose disk serial in virDomainGetGuestInfo()
2206 The ``virDomainGetGuestInfo()`` reports disk serial number among with other
2211 * qemu: Fix crash of libvirt on full block pull of a disk
2213 When the persistent definition contains a compatible disk (meaning the
2214 definition of the running and persistent config match) a block pull job
2215 would leave a dangling pointer in the config definition which resulted
2218 * qemu: Use proper job cancelling command
2220 Libvirt's API contract for aborting a block copy job in 'ready' state
2221 declares that the destination image of the copy will contain a consistent
2222 image of the disk from the time when the block job was aborted. This
2223 requires that libvirt uses the proper cancelling qemu command to ensure
2224 that the data is consistent which was not the case.
2226 * qemu: Don't attempt storage migration when there are no migratable disks
2228 Due to a logic bug introduced in the previous release libvirt would attempt
2229 to migrate disks in case when no disks are selected/eligible for migration.
2231 * qemu: Fix very rare race when two block job 'ready' events are delivered
2233 In certain high-load scenarios, qemu might deliver the 'ready' event twice
2234 and if it's delivered when pivoting to the destination during a block copy
2235 job, libvirt would get confused and execute the code as if the job were
2238 * lxc: Fix container destroy with CGroupsV2
2240 When an LXC container was started and the host used CGroupsV2 it might have
2241 had created nested controllers under the container's scope. Libvirt was
2242 unaware and thus destroying the container failed with a cryptic error:
2243 ``failed to get cgroup backend for 'pathOfController'``. The CGroup removal
2244 code was reworked and is now capable of dealing with such scenario.
2246 * bash-completion: Fix argument passing to $1
2248 Due to a bug in bash completion script, the auto completion did not work
2249 properly when a connection URI or read only flag were specified on
2250 ``virsh`` or ``virt-admin`` command line.
2258 * qemu: Implement domain memory dirty rate calculation API
2260 New API ``virDomainStartDirtyRateCalc()`` and virsh command
2261 ``domdirtyrate-calc`` are added to start calculating a live domain's
2264 * qemu: Support reporting memory dirty rate stats
2266 The memory dirty rate stats can be obtained through ``virsh domstats
2267 --dirtyrate`` via the virConnectGetAllDomainStats API.
2269 * qemu: Full disk backups via ``virDomainBackupBegin``
2271 The qemu hypervisor driver now allows taking full disk backups via the
2272 ``virDomainBackupBegin`` API and the corresponding virsh wrapper.
2274 In future releases the feature will be extended to also support incremental
2275 backups (where only the difference since the last backup is copied) when
2276 qemu adds the required functionality.
2278 * Add support for audio backend specific settings
2280 With this release a new ``<audio/>`` element is introduced that allows
2281 users to configure audio output for their guests.
2285 * qemu: Compatibility with QEMU 6.0 for certain hot-(un)-plug operations
2287 Libvirt 7.2.0 is required for compatibility with the upcoming QEMU 6.0
2288 release for hotplug and hotunplug of certain devices and helpers, such as
2289 iothreads, chardevs, RNG devices, disks with secret, ...
2291 * qemu: Various improvements to embedded mode
2293 Embedded mode for the QEMU driver, as well as the ``virt-qemu-run`` tool
2294 saw improvements in handling of domain life cycle, temporary directories
2295 creation (important when using disk secrets) and other minor fixes.
2297 * Documentation of split daemon related config files
2299 Split daemons read configuration files upon their start. These were never
2304 * Check host CPU for forbidden features
2306 CPU feature policy did not work as expected with ``host-passthrough`` and
2307 features supported by physical host. CPU features were not filtered out
2308 when ``@check`` was set to ``full``.
2310 * Fix virNetworkUpdate() to work with split daemons
2312 Due to a bug in our code, virNetworkUpdate() did not work with split daemon
2313 unless management application connected to virtnetworkd directly.
2315 * qemu: increase locked memory limit when a vDPA device is present
2317 Just like VFIO devices, vDPA devices may need to have all guest memory
2318 pages locked/pinned in order to operate properly. These devices are now
2319 included when calculating the limit for memory lock.
2321 * Don't log error if SRIOV PF has no associated netdev
2323 Some SRIOV PFs don't have a netdev associated with them in which case
2324 libvirtd reported an error and refused to start. This is now fixed.
2326 * qemu: Only raise memlock limit if necessary
2328 Attempting to set the memlock limit might fail if we're running
2329 in a containerized environment where ``CAP_SYS_RESOURCE`` is not
2330 available, and if the limit is already high enough there's no
2331 point in trying to raise it anyway.
2333 * Restore security context of swtpm.log
2335 If a guest with emulated TPM was started and the daemon was restarted
2336 afterwards, the security context of the per-domain ``swtpm.log`` file was
2337 not restored on domain shutdown leaving it unable to be started again.
2339 * virtlogd|virtlockd: Fixed crash when upgrading the daemons in-place
2341 A bug preventing the in-place upgrade of ``virtlogd`` and ``virtlockd``
2342 daemons was fixed, so they can again be upgraded without dropping the log
2343 file descriptors or locks on files.
2351 * Implement Apple Silicon support
2353 libvirt now runs on the ARM-based Apple Silicon Macs.
2357 * Introduce virtio-pmem ``<memory/>`` model
2359 The virtio-pmem is a virtio variant of NVDIMM and just like NVDIMM
2360 virtio-pmem also allows accessing host pages bypassing guest page cache.
2362 * Introduce ``<boot order/>`` for ``<filesystem>``
2364 Booting is possible from virtiofs filesystems. Introduce an option
2365 to control the boot order, like we do for other bootable devices.
2367 * hyperv: implement new APIs
2369 The ``virDomainUndefine()``, ``virDomainUndefineFlags()``,
2370 ``virDomainDefineXML()``, ``virDomainAttachDevice()``, and
2371 ``virDomainAttachDeviceFlags()``, ``virConnectListAllNetworks()``,
2372 ``virConnectNumOfNetworks()``, ``virNetworkLookupByName()``,
2373 ``virNetworkLookupByUUID()``, ``virConnectNumOfDefinedNetworks()``,
2374 ``virConnectListDefinedNetworks()``, ``virNetworkGetAutostart()``,
2375 ``virNetworkIsActive()``, ``virNetworkIsPersistent()``,
2376 ``virNetworkGetXMLDesc()``, and ``virDomainScreenshot()``, APIs have been
2377 implemented in the Hyper-V driver.
2379 * Support <teaming> element in plain <hostdev> devices
2381 This is useful when libvirt doesn't have the privileges necessary
2382 to set the hostdev device's MAC address (which is a necessary
2383 part of the alternate <interface type='hostdev'>).
2385 * Introduce ``<disk type='vhostuser'>`` support
2387 Introduces support for QEMU vhost-user-blk device that can be used
2388 to access storage exported via the vhost-user protocol by daemons such
2389 as the ``qemu-storage-daemon``.
2393 * qemu: Fix disk quiescing rollback when creating external snapshots
2395 If the qemu guest agent call to freeze filesystems failed when creating
2396 an external snapshot with ``VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE`` flag the
2397 filesystems would be unconditionally thawed. This could cause problems when
2398 the filesystems were frozen by an explicit call to ``virDomainFSFreeze``
2399 since the guest agent then rejects any further freeze attempts once are
2400 filesystems frozen, an explicit freeze followed by a quiesced snapshot
2401 would fail and thaw filesystems.
2403 Users are also encouraged to use ``virDomainFSFreeze/Thaw`` manually instead
2404 of relying on ``VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE`` if they need finer
2407 * cgroups: Fix how we setup and configure cgroups on hosts with systemd
2409 When libvirt is running on host with systemd we register every VM with
2410 machined which creates the VM root cgroup for us as well. Before this fix
2411 we were directly modifying files in the VM root cgroup which was incorrect
2412 because all the files are managed by systemd. The implication was that any
2413 change done by libvirt to cgroup attributes supported by systemd could be
2414 removed which happens for example by running ``systemctl daemon-reload``.
2416 To fix the issue libvirt now uses DBus calls for some of the cgroup
2417 attributes that distribute the resources proportionally to the cgroup
2418 siblings and for the rest we have a new sub-cgroup that libvirt can
2421 For more details why this is necessary see
2422 `systemd cgroup <https://systemd.io/CGROUP_DELEGATION/>`_ documentation.
2424 * qemu: Fix swtpm device with aarch64
2426 The TPM TIS device name for x86 is ``tpm-tis``, whereas for aarch64 it is
2427 ``tpm-tis-device``. Fix the use of TPM TIS device with aarch64 by using
2428 the proper device name when building the QEMU command line.
2430 * libxl: Fix domain shutdown
2432 Commit fa30ee04a2 introduced the possibility of a race between the
2433 shutdown and death threads used to process domain shutdown and death
2434 events from libxl. On normal domain shutdown the shutdown thread handles
2435 all aspects of shutting down and cleaning up the domain. The death
2436 thread is only used to handle out-of-band domain destruction and is
2437 inhibited when domain shutdown is under libvirt's control. The race is
2438 avoided by also inhibiting the death thread when libvirt starts the
2445 * **Project governance**
2447 * Formal handover of release tarball signing
2449 Starting from libvirt-6.6 the release tarballs are signed by Jiřà Denemark.
2450 Releases starting with 7.0 contain a note from the previous maintainer
2451 Daniel Veillard officially handing over the signing of packages so that the
2452 transition can be verified.
2456 * nodedev: Add node device driver support for AP devices
2458 Add support for detecting and listing Adjunct Processor(AP) cards, AP
2459 queues and AP matrix devices (which are capable of MDEV) of a KVM host
2460 system in libvirt node device driver with correct object relationships.
2462 * qemu: Allow control of ``qcow2`` metadata cache
2464 In specific usecases such as when massive storage images are used it's
2465 possible to achieve better performance by increasing the metadata cache
2466 size. The new knob allows advanced users setting the size according to
2467 qemu's documentation to suit their image.
2469 * conf: Add support for keeping TPM emulator state
2471 Currently, swtpm TPM state file is removed when a transient domain is
2472 powered off or undefined. Add per-TPM emulator option ``persistent_state``
2473 for keeping TPM state.
2475 * cpu_map: Add Snowridge CPU model
2477 It's supported in QEMU 4.1 and newer.
2479 * qemu: Add support for NFS disk protocol
2481 Implement support for the 'nfs' native protocol driver in the qemu driver.
2485 * qemu: Discourage users from polling ``virDomainGetBlockJobInfo`` for block
2488 Document that waiting for events is a more robust solution.
2490 * secret: Relax XML schema for the ``usage`` name of a ``secret``
2492 Various bits of documentation of how to use libvirt with RBD volumes used
2493 an usage name which would not pass the XML validation. Relax the requirement
2494 to make such XMLs valid.
2496 * virnetdevopenvswitch: Various improvements
2498 The code that handles ``<interface type='vhostuser'/>`` was given various
2499 improvements. So far, libvirt assumed vhostuser interfaces are handled
2500 exclusively by OpenVSwitch and refused to start a guest if it was not so.
2501 Now a guest can be started successfully even if the interface is created by
2502 some other tool (e.g. ``dpdk-testpmd``). Also, the code that detects the
2503 interface name was adapted to new versions of OpenVSwitch and thus can
2504 detect name more reliably.
2506 * qemu: Report guest disks information in ``virDomainGetGuestInfo``
2508 Libvirt is now able to report disks and filesystems from the guest's
2509 perspective (using guest agent). And with sufficiently new guest agent
2510 (5.3.0 or newer) the API also handles disks on CCW bus.
2514 * qemu: Fix logic bug in inactive snapshot deletion
2516 This release fixes a bug introduced in libvirt-6.9 where libvirt's
2517 snapshot metadata would not be deleted on successful snapshot deletion.
2519 * qemu: Fix VMs with ``<iotune>`` on an empty cdrom
2521 Specifying ``<iotune>`` for an empty cdrom would prevent the VM from
2522 starting as qemu doesn't accept the tuning for an empty drive. We now
2523 postpone setting the parameters until a new media is inserted.
2525 * Avoid taking extra host memory when launching pSeries guests
2527 Under certain conditions, pSeries guests were being launched with more
2528 RAM than it was specified in the domain XML by the user. New pSeries
2529 domains created with libvirt 7.0.0 will always launch with the right
2530 amount of initial memory. Existing guests that migrate from an older
2531 libvirt version to 7.0.0 will not be affected by this change.
2533 * qemu: Don't cache NUMA caps
2535 ``virsh capabilities`` contains ``<topology/>`` section which reports NUMA
2536 topology among with amount of free hugepages per each NUMA node. However,
2537 these amounts were not updated between calls.
2539 * networkGetDHCPLeases: Handle leases with infinite expiry time
2541 Since libvirt-6.3.0 it is possible to configure expiry time for DHCP
2542 leases. If the expiry time was infinite then ``virsh net-dhcp-leases``
2543 and NSS plugins refused to work.
2545 * qemu: Don't prealloc mem for real NVDIMMs
2547 If a real life NVDIMM is assigned to a guest via ``<memory model='nvdimm'/>``
2548 then QEMU is no longer instructed to preallocate memory
2549 for it. This prevents unnecessary wear on the NVDIMM.
2551 * network: Introduce mutex for bridge name generation
2553 When new libvirt network is defined or created and the input XML does not
2554 contain any bridge name, libvirt generates one. However, it might have
2555 happened that the same name would be generated for different networks if
2556 two or more networks were defined/created at once.
2559 v6.10.0 (2020-12-01)
2560 ====================
2564 * qemu: Enable client TLS certificate validation by default for ``chardev``,
2565 ``migration``, and ``backup`` servers.
2567 The default value if qemu.conf options ``chardev_tls_x509_verify``,
2568 ``migrate_tls_x509_verify``, or ``backup_tls_x509_verify`` are not
2569 specified explicitly in the config file and also the
2570 ``default_tls_x509_verify`` config option is missing are now '1'. This
2571 ensures that only legitimate clients access servers, which don't have any
2572 additional form of authentication.
2574 * qemu: Introduce "migrate_tls_force" qemu.conf option
2576 The ``migrate_tls_force`` configuration option allows administrators to
2577 always force connections used for migration to be TLS secured as if the
2578 ``VIR_MIGRATE_TLS`` flag had been used.
2582 * qemu: Implement OpenSSH authorized key file management APIs
2584 New APIs (``virDomainAuthorizedSSHKeysGet()`` and
2585 ``virDomainAuthorizedSSHKeysSet()``) and virsh commands
2586 (``get-user-sshkeys`` and ``set-user-sshkeys``) are added to manage
2587 authorized_keys SSH file for user.
2589 * hyperv: implement new APIs
2591 The ``virDomainGetMaxMemory()``, ``virDomainSetMaxMemory()``,
2592 ``virDomainGetSchedulerType()``, ``virDomainGetSchedulerParameters()``,
2593 ``virDomainGetSchedulerParametersFlags()``, ``virDomainGetVcpus()``,
2594 ``virDomainGetVcpusFlags()``, ``virDomainGetMaxVcpus()``,
2595 ``virDomainSetVcpus()``, and ``virDomainSetVcpusFlags()`` APIs have been
2596 implemented in the Hyper-V driver.
2598 * qemu: Add 'fmode' and 'dmode' options for 9pfs
2600 Expose QEMU's 9pfs 'fmode' and 'dmode' options via attributes on the
2601 'filesystem' node in the domain XML. These options control the creation
2602 mode of files and directories, respectively, when using accessmode=mapped.
2603 It requires QEMU 2.10 or above.
2605 * qemu: support kvm-poll-control performance hint
2607 Implement the new KVM feature 'poll-control' to set this performance hint
2608 for KVM guests. It requires QEMU 4.2 or above.
2612 * virsh: Support network disks in ``virsh attach-disk``
2614 The ``virsh attach-disk`` helper command which simplifies attaching of disks
2615 without the need for the user to formulate the disk XML manually now
2616 supports network-backed images. Users can specify the protocol and host
2617 specification with new command line arguments. Please refer to the man
2618 page of virsh for further information.
2622 * remote: fixed performance regression in SSH tunnelling
2624 The ``virt-ssh-helper`` binary introduced in 6.8.0 had very
2625 poor scalability which impacted libvirt tunnelled migration
2626 and storage volume upload/download in particular. It has been
2627 updated and now has performance on par with netcat.
2629 * **Removed features**
2631 * hyperv: removed support for the Hyper-V V1 WMI API
2633 This drops support for Windows Server 2008R2 and 2012.
2634 The earliest supported version is now Windows 2012R2.
2642 * nodedev: Add support for channel subsystem (CSS) devices on S390
2644 A CSS device is represented as a parent device of a CCW device.
2645 This support allows to create vfio-ccw mediated devices with
2646 ``virNodeDeviceCreateXML()``.
2648 * qemu: Implement memory failure event
2650 New event is implemented that is emitted whenever a guest encounters a
2653 * qemu: Implement support for ``<transient/>`` disks
2655 VMs based on the QEMU hypervisor now can use ``<transient/>`` option for
2656 local file-backed disks to configure a disk which discards changes made to
2657 it while the VM was active.
2659 * hyperv: implement new APIs
2661 The ``virConnectGetCapabilities()``, ``virConnectGetMaxVcpus()``,
2662 ``virConnectGetVersion()``, ``virDomainGetAutostart()``,
2663 ``virDomainSetAutostart()``, ``virNodeGetFreeMemory()``,
2664 ``virDomainReboot()``, ``virDomainReset()``, ``virDomainShutdown()``, and
2665 ``virDomainShutdownFlags()`` APIs have been implemented in the Hyper-V
2668 * bhyve: implement virtio-9p filesystem support
2670 Implement virito-9p shared filesystem using the ``<filesystem/>`` element.
2672 * qemu: Add support for vDPA network devices.
2674 VMs using the QEMU hypervisor can now specify vDPA network devices
2675 using ``<interface type='vdpa'>``. The node device APIs also now
2676 list and provide XML descriptions for vDPA devices.
2678 * cpu_map: Add EPYC-Rome CPU model
2680 It's supported in QEMU 5.0.0 and newer.
2682 * cpu: Add a flag for XML validation in CPU comparison
2684 The ``virConnectCompareCPU`` and ``virConnectCompareHypervisorCPU`` API
2685 now support the ``VIR_CONNECT_COMPARE_CPU_VALIDATE_XML`` flag, which
2686 enables XML validation. For virsh, this feature is enabled by passing
2687 the ``--validate`` option to the ``cpu-compare`` and
2688 ``hypervisor-cpu-compare`` subcommands.
2690 * qemu: Introduce virtio-balloon free page reporting feature
2692 Introduce the optional attribute ``free-page-reporting`` for virtio
2693 memballoon device. It enables/disables the ability of the QEMU virtio
2694 memory balloon to return unused pages back to the hypervisor. QEMU 5.1
2695 and newer support this feature.
2699 * qemu: Make 'cbitpos' & 'reducedPhysBits' attrs optional
2701 Libvirt probes the underlying platform in order to fill in these SEV
2702 attributes automatically before launching a guest.
2704 * util: support device stats collection for SR-IOV VF hostdev
2706 For SR-IOV VF hostdevs, libvirt now supports retrieving device traffic
2707 stats via the ``virDomainInterfaceStats`` API and ``virsh domifstat``.
2709 * logging: Allow disabling log rollover
2711 Set ``max_len=0`` in ``virtlogd.conf`` to disable log rollover.
2713 * qemu: Set noqueue qdisc for TAP devices
2715 Set ``noqueue`` instead of the former ``pfifo_fast`` queue discipline
2716 for TAP devices. It will avoid needless cost of host CPU cycles and
2717 thus improve performance.
2719 * qemu: virtiofs can be used without NUMA nodes
2721 Virtiofs is supported for the VM without NUMA nodes but configured with
2726 * hyperv: ensure WQL queries work in all locales
2728 Relying on the "Description" field caused queries to fail on non-"en-US"
2729 systems. The queries have been updated to avoid using localized strings.
2731 * rpc: Fix ``virt-ssh-helper`` detection
2733 libvirt 6.8.0 failed to correctly detect the availability of the new
2734 ``virt-ssh-helper`` command on the remote host, and thus always used the
2735 fallback instead; this has now been fixed.
2743 * qemu: double free in qemuAgentGetInterfaces() in qemu_agent.c
2745 Clients connecting to the read-write socket with limited ACL permissions
2746 may be able to crash the libvirt daemon, resulting in a denial of service,
2747 or potentially escalate their privileges on the system. CVE-2020-25637.
2751 * xen: Add ``writeFiltering`` attribute for PCI devices
2753 By default Xen filters guest writes to the PCI configuration space of a
2754 PCI hostdev, which may cause problems for some devices. The ``writeFiltering``
2755 attribute of the device's ``<source>`` element can be used to disable the
2756 filtering and allow all guest writes to the configuration space.
2758 * bhyve: Support setting the framebuffer resolution
2760 Libvirt can now set the framebuffer's "w" and "h" parameters
2761 using the ``resolution`` element.
2763 * bhyve: Support VNC password authentication
2765 Libvirt can now probe whether the bhyve binary supports
2766 VNC password authentication. In case it does, a VNC password
2767 can now be passed using the ``passwd`` attribute on
2768 the ``<graphics>`` element.
2770 * remote: ``virt-ssh-helper`` replaces ``nc`` for SSH tunnelling
2772 Libvirt now provides a ``virt-ssh-helper`` binary on the server
2773 side. The libvirt remote client will use this binary for setting
2774 up an SSH tunnelled connection to hosts. If not present, it will
2775 transparently fallback to the traditional ``nc`` tunnel. The new
2776 binary makes it possible for libvirt to transparently connect
2777 across hosts even if libvirt is built with a different installation
2778 prefix on the client vs server. It also enables remote access to
2779 the unprivileged per-user libvirt daemons (e.g. using a URI such as
2780 ``qemu+ssh://hostname/session``). The only requirement is that
2781 ``virt-ssh-helper`` is present in ``$PATH`` of the remote host.
2783 * esx: implement few APIs
2785 The ``virConnectListAllNetworks()``, ``virDomainGetHostname()``, and
2786 ``virDomainInterfaceAddresses()`` (only for
2787 ``VIR_DOMAIN_INTERFACE_ADDRESSES_SRC_AGENT`` source) APIs were implemented
2792 * qemu: Allow migration over UNIX sockets
2794 QEMU migration can now be performed completely over UNIX sockets. This is
2795 useful for containerised scenarios and can be used in both peer2peer and
2798 * dbus: Use GLib implementation instead of libdbus
2800 Adopting GLib DBus implementation simplifies our code as libdbus provides
2801 low-level APIs where we had to have a lot of helper functions. With this
2802 change we also remove dependency on libdbus and possibly fix all the DBus
2803 related libvirtd crashes seen over the time.
2805 * Re-introduce NVDIMM auto-alignment for pSeries Guests
2807 The auto-alignment logic was removed in v6.7.0 in favor of requiring the
2808 size provided by the user to be already aligned; however, this had the
2809 unintended consequence of breaking some existing guests. v6.8.0 restores
2810 the previous behavior with an improvement: it also reflects the auto-aligned
2811 value in the domain XML.
2813 * qemu: Preserve qcow2 cluster size after external snapshots
2815 The new overlay image which is installed on top of the current chain when
2816 taking an external snapshot now preserves the cluster size of the original
2817 top image to preserve any performance tuning done on the original image.
2821 * qemu: Various (i)SCSI backed hostdev fixes
2823 (i)SCSI backed hostdevs now work again with an arbitrarily long
2824 user-specified device alias and also honor the 'readonly' property after a
2827 * **Removed features**
2829 * node_device: Remove HAL node device backend
2831 HAL is deprecated on all supported OS so there is no need to keep it
2832 in libvirt. udev backend is used on Linux OSes and devd can be eventually
2833 implemented as replacement for FreeBSD.
2839 * **Packaging changes**
2841 * Libvirt switch to Meson build system
2843 Libvirt abandoned autotools and switched to Meson build system.
2847 * qemu: Add support for initiator IQN configuration for iSCSI hostdevs
2849 Similarly to iSCSI ``<disk>`` users can use an ``<initiator>`` element
2850 inside ``<hostdev>`` with the same format to configure the ``IQN`` value
2851 used by the qemu initiator when connecting to an iSCSI target.
2853 * xen: Add support for device model command-line passthrough
2855 Xen supports passing arbitrary arguments to the QEMU device model using
2856 the ``device_model_args`` setting in xl.cfg(5). The libvirt xen driver now
2857 supports this using ``<xen:commandline/>`` XML extensions.
2859 * shmem: Add support for shmem-{plain, doorbell} ``role`` option
2861 The ``role`` attribute controls how the domain behaves on migration. With
2862 ``role=master``, the guest will copy the shared memory on migration to
2863 the destination host. With ``role=peer``, the migration is disabled.
2865 * bhyve: Sound device support
2867 This feature allows to configure guest sound device using
2868 the ``<sound>`` element, and map it to the host sound device using
2869 the ``<audio>`` element.
2873 * Allow sparse streams for block devices
2875 Sparse streams (e.g. ``virsh vol-download --sparse`` or ``virsh vol-upload
2876 --sparse``) now handle if one of the stream ends is a block device.
2878 * Remove NVDIMM auto-alignment for pSeries Guests
2880 This feature was introduced in libvirt v6.2.0 as part of the overall
2881 NVDIMM support for pSeries guests. The idea was to relieve the user
2882 from knowing ppc64 alignment details, but the end result is that we
2883 ended up with inconsistencies between domain XML and actual NVDIMM
2884 size the guest is using. To promote consistency between domain XML
2885 and the guest, unaligned NVDIMM sizes for pSeries guests will now be
2886 forbidden and no size auto-alignment will be made. Instead, libvirt will
2887 suggest an aligned round up size for the user.
2889 * apparmor: Several improvements
2891 Add support for virtiofs filesystem and allow QEMU to load old
2892 shared objects after upgrade.
2896 * virdevmapper: Deal with kernels without DM support
2898 In the previous release libvirt dropped libdevmapper in favor of its own
2899 implementation. However, it failed to deal correctly with kernels that
2900 either don't have device mapper enabled or where the dm-mod module is not
2901 loaded yet. This is now fixed.
2903 * resctrl: Use exclusive lock for /sys/fs/resctrl
2905 When two or more domains were attempted to start at once, due to a bug in
2906 implementation, resctrl was not locked properly and thus threads did not
2907 mutually exclude with each other resulting in not setting requested
2910 * mdev: Fix daemon crash when reattaching mdevs on assignment conflict
2912 If there's a list of mdevs to be assigned to a domain, but one of them (NOT
2913 the first) is already assigned to a different domain then libvirtd would
2914 crash. This is now fixed.
2916 * Fix logic in setting COW flag on btrfs
2918 When COW is not explicitly requested to be disabled or enabled, then
2919 libvirt should do nothing on non-BTRFS file systems.
2921 * Avoid crash due to race in glib event loop code
2923 Libvirt switched to glib event loop in 6.1.0 but it was also tickling a bug
2924 in glib code leading to the daemon crash. Libvirt way of calling glib was
2925 changed so the daemon crashes no more.
2927 * virdevmapper: Handle kernel without device-mapper support
2929 In the previous release, Libvirt dropped libdevmapper in favor of its own
2930 implementation. But the implementation did not handle kernels without
2931 device-mapper support. This is now fixed.
2933 * remove autogenerated macvtap names from migration XML
2935 Autogenerated macvtap device names were being left in the
2936 migration XML, which could result in libvirt erroneously deleting
2937 the macvtap device of a different guest in the aftermath of
2938 failing to restart the guest on the destination host. Removing the
2939 autogenerated names avoids this.
2947 * Allow configuring of ACPI NUMA HMAT
2949 Libvirt allows configuring ACPI Heterogeneous Memory Attribute Table to
2950 hint software running inside the guest on optimization.
2952 * esx: Add a ``type`` attribute for mac addresses.
2954 This attribute allows (when set to ``static``) ignoring VMWare checks of the
2955 MAC addresses that would generate a new one if they were in its OUI
2958 * conf: add control over COW for storage pool directories
2960 The storage pool code now attempts to disable COW by default on btrfs, but
2961 management applications may wish to override this behaviour. This is now
2962 possible via new ``cow`` element.
2966 * esx: Change the NIC limit for recent virtualHW versions
2968 Specifying a virtualHW version greater or equal to 7 (ESXi 4.0) will allow
2969 you to use up to 10 NICs instead of 4 as it was previously.
2971 * qemu: Support encrypted TLS keys for NBD disks
2973 The secret key used for disks can now be encrypted similarly to TLS keys
2974 used for migration, chardev and others.
2976 * qemu: ``VIR_DOMAIN_EVENT_ID_BLOCK_THRESHOLD`` can now be registered for ``<mirror>``
2978 The event can now be used also for block copy destinations by using the
2979 index of the ``<mirror>`` image.
2981 * qemu: consider available CPUs in ``vcpupin/emulatorpin`` output
2983 This patch changes the default bitmap of ``vcpupin`` and ``emulatorpin``,
2984 in the case of domains with static vcpu placement, all available CPUs
2985 instead of all possible CPUs are returned making these APIs consistent with
2986 the behavior of ``vcpuinfo``.
2990 * virdevmapper: Don't use libdevmapper to obtain dependencies
2992 When building domain's private ``/dev`` in a namespace, libdevmapper was
2993 consulted for getting full dependency tree of domain's disks. However, this
2994 meant that libdevmapper opened ``/dev/mapper/control`` which wasn't closed
2995 and was leaked to QEMU. CVE-2020-14339
2997 * qemu: Report correct ``index`` in ``VIR_DOMAIN_EVENT_ID_BLOCK_THRESHOLD``
2999 Starting from libvirt 5.10 with QEMU 4.2 the
3000 ``VIR_DOMAIN_EVENT_ID_BLOCK_THRESHOLD`` event would report incorrect device
3001 ``index`` when reported for an image from the backing chain of a disk.
3003 * qemu: Don't fail active layer block commit or block copy in certain cases
3005 Starting from libvirt-6.5 an active layer block commit or a block copy could
3006 fail if the same destination was used more than once.
3008 * qemu: Don't change ownership of restore file
3010 When restoring a domain from a file, Libvirt no longer changes its ownership.
3012 * qemu: Set SPAPR TPM default to 2.0 and prevent 1.2 choice
3014 The firmware (SLOF) on QEMU for ppc64 does not support TPM 1.2, so prevent
3015 the choice of TPM 1.2 when the SPAPR device model is chosen and use a
3016 default of '2.0' (TPM 2) for the backend.
3018 * qemu: Do not set ``//cpu/@migratable`` for running domains
3020 Libvirt release of 6.4.0 started to fill the default value for
3021 ``//cpu/@migratable`` attribute according to QEMU support. However, active
3022 domains either have the migratable attribute already set or they were
3023 started with older Libvirt which doesn't support the attribute.
3031 * Allow firmware blobs configuration
3033 QEMU offers a way to tweak how firmware configures itself
3034 and/or provide new configuration blobs. New ``<sysinfo/>``
3035 type is introduced that will hold these new blobs.
3036 It's possible to either specify new value as a string or
3037 provide a filename which contents then serve as the value.
3039 * nodedev: Add ability to create mediated devices
3041 Mediated devices can now be created with ``virNodeDeviceCreateXML()``. This
3042 functionality requires the ``mdevctl`` utility to be installed. The XML
3043 schema for node devices was expanded to support attributes for mediated
3046 * QEMU: add TPM Proxy device support
3048 libvirt can now create guests using a new device type called
3049 "TPM Proxy". The TPM Proxy connects to a TPM Resource Manager
3050 present in the host, enabling the guest to run in secure virtual
3051 machine mode with the help of an Ultravisor. Adding a TPM Proxy to
3052 a pSeries guest brings no security benefits unless the guest is
3053 running on a PPC64 host that has Ultravisor and TPM Resource Manager
3054 support. Only one TPM Proxy is allowed per guest. A guest using
3055 a TPM Proxy device can instantiate another TPM device at the same
3056 time. This device is supported only for pSeries guests via the new
3057 'spapr-tpm-proxy' model of the TPM 'passthrough' backend.
3059 * virhook: Support hooks placed in several files
3061 Running all scripts from directory /etc/libvirt/hooks/<driver>.d in
3062 alphabetical order. Hook script in old place will be executed
3063 as first for backward compatibility.
3065 * qemu: Add support for migratable host-passthrough CPU
3067 QEMU 2.12 made it possible for guests to use a migration-friendly
3068 version of the host-passthrough CPU. This feature is now exposed by
3073 * network: Support NAT with IPv6
3075 It's now possible to use ``<nat ipv6="yes"/>`` in a libvirt network.
3077 * qemu: Auto-fill NUMA information for incomplete topologies
3079 If the NUMA topology is not fully described in the guest XML, libvirt
3080 will complete it by putting all unspecified CPUs in the first NUMA node.
3081 This is only done in the QEMU binary itself supports disjointed CPU
3082 ranges for NUMA nodes.
3084 * qemu: Assign hostdev-backed interfaces to PCIe slots
3086 All SR-IOV capable devices are PCIe, so when their VFs are assigned to
3087 guests they should end up in PCIe slots rather than conventional PCI ones.
3091 * qemu: fixed crash in ``qemuDomainBlockCommit``
3093 This release fixes a regression which was introduced in libvirt v6.4.0
3094 where libvirtd always crashes when a block commit of a disk is requested.
3096 * qemu: fixed zPCI address auto generation on s390
3098 Removes the correlation between the zPCI address attributes uid and fid.
3099 Fixes the validation and autogeneration of zPCI address attributes.
3101 * qemu: Skip pre-creation of NVMe disks during migration
3103 libvirt has no way to create NVMe devices on the target host, so it now
3104 just makes sure they exist and let the migration proceed in that case.
3112 * qemu: Add support for pvscsi controllers
3114 pvscsi is the VMware paravirtualized SCSI controller, which has been
3115 supported in QEMU for a number of years.
3117 * cpu: Report model information for ARM CPUs
3119 ``virsh capabilities`` will now include information about the host CPU when
3120 run on ARM machines.
3122 * qemu: support network interface downscript
3124 QEMU has the ability to run a script when a NIC is brought up and down.
3125 Libvirt only enables use of the up script. Now add support for postscript
3126 when NIC is down/detached.
3130 * qemu: stricter validation for disk type='lun'
3132 The 'lun' type is meant for SCSI command passthrough, which can't be
3133 achieved if qemu's block layer features are used. Disk type='lun' is now
3134 allowed only when the format is 'raw' and no other block layer features are
3137 * qemu: auto-fill of incomplete NUMA topologies
3139 Domains with incomplete NUMA topologies, where the sum of vCPUs in all NUMA
3140 cells is less than the total of vCPUs, will get their first NUMA cell to
3141 be auto-filled with the remaining vCPUs. This behavior reproduces what QEMU
3142 already does in these cases. Users are encouraged to provide complete NUMA
3143 topologies to avoid unexpected changes in the domain XML.
3145 * Cooperlake x86 CPU model is added
3149 * qemu: fixed regression in network device hotplug with new qemu versions
3151 Starting from QEMU-5.0 it's required to conform to strict schema when
3152 hotplugging network devices. Libvirt didn't conform to the schema so in
3153 versions prior to 6.4.0 network device hotplug fails in certain cases. This
3154 version fixes it and adds stricter testing to prevent further issues.
3156 * remote: Look up libxl driver correctly
3158 This makes ``xen://`` connection URIs usable in split daemon mode.
3160 * systemd: Start libvirtd after firewalld/iptables services
3162 This solves an issue where iptables rules and chains created by libvirtd
3163 would get removed by a service started after it.
3165 * network: Re-create iptables chains on firewalld restart
3167 firewalld resets all iptables rules and chains on restart, and this
3168 includes deleting those created by libvirt.
3170 * qemu: reject readonly attribute for virtiofs
3172 virtiofs does not yet support read-only shares.
3180 * qemu: support disabling hotplug/unplug of PCIe devices
3182 libvirt can now set the "hotplug" option for pcie-root-ports and
3183 pcie-switch-downstream-ports, which can be used to disable hotplug/unplug
3184 of devices from these ports (default behavior is for these controllers to
3185 accept all hotplug/unplug attempts, but this is often undesirable).
3187 * vbox: added support for version 6.0 and 6.1 APIs
3189 libvirt can now support use of the VirtualBox 6.0 and 6.1 APIs. This is
3190 compile tested only, so we are looking for feedback from users on how well
3191 it works in practice.
3193 * xen: Add support for 'e820_host' hypervisor feature
3195 ``e820_host`` is a Xen-specific option only available for PV guests. When
3196 enabled it provides the guest with a virtual e820 memory map based on the
3197 host one. It must be enabled to allow hotplugging PCI devices to PV guests,
3198 particularly when memory ballooning is enabled.
3200 * xen: Add support for 'passthrough' hypervisor feature
3202 ``passthrough`` is a Xen-specific option new to Xen 4.13 that enables PCI
3203 passthrough for guests. It must be enabled to allow hotplugging PCI
3206 * qemu: support async IO mode 'io_uring'
3208 Linux 5.1 introduces a fast and efficient async IO interface io_uring, then
3209 qemu implements that in version 5.0. It could be used in file, host device
3210 and host cdrom backend by ``io='io_uring'`` of disk XML.
3212 * Lease time option included for network DHCP settings
3214 Users can now configure expiry time for leases for networks where libvirt
3215 manages DHCP. The time can be specified for whole range and/or fine tuned
3216 per individual host.
3218 * qemu: Implement pSeries Spectre mitigation features
3220 Users can now setup the following capabilities of pSeries guests: CFPC
3221 (Cache Flush on Privilege Change), SBBC (Speculation Barrier Bounds
3222 Checking) and IBS (Indirect Branch Speculation).
3224 * qemu: Add support for virtio packed option
3226 The ``packed`` attribute controls if QEMU should try to use packed
3227 virtqueues. Possible values are ``on`` or ``off``.
3231 * qemu: Allow checkpoint redefine for offline VMs
3233 Skip the liveness and capability checks when redefining checkpoints as we
3234 don't need QEMU interactions to update the metadata.
3236 * daemons: Improve timeout handling
3238 Daemons now support ``--timeout 0`` which suppresses daemon killing after
3239 given time of inactivity.
3241 * qemu: Add support for 'multidevs' option
3243 This option prevents misbehaviours on guest if a QEMU 9pfs export contains
3244 multiple devices, due to the potential file ID collisions this otherwise
3249 * qemu: Various embed driver fixes
3251 When using shared resources from embed driver (e.g. hugepages, machined,
3252 etc.) libvirt now generates an unique handler that is not conflicting with
3253 other embed drivers or system or session daemons.
3255 * cpu: Distinguish Cascadelake-Server from Skylake-Server
3257 Libvirt now properly detects Cascadelake-Server and Skylake-Server
3258 processors which differ only in stepping.
3260 * qemu: Fix domain restore from a block device
3262 When using namespaces, libvirt was unable to restore a domain from a block
3263 device because libvirt tried to relabel the device inside the namespace
3264 while QEMU was given FD to the block device in the host.
3266 * node_device_udev: Handle move events
3268 Libvirt now handles ``move`` event which is emitted on a NIC rename.
3270 * qemu: Fix capabilities probing with TCG
3272 Libvirt no long assumes TCG is always available. It now detects whether
3273 QEMU supports TCG and reports it accordingly.
3275 * **Removed features**
3277 * vbox: removed support for version 5.0 and 5.1 APIs
3279 libvirt no longer supports use of VirtualBox 5.0 and 5.1 since these
3280 versions reached their end of life on 2017/05 and 2018/04 respectively.
3288 * qemu: NVDIMM support for pSeries guests
3290 QEMU 5.0 implements NVDIMM memory support for pSeries guests. This is done
3291 by adding an 'uuid' element in the memory XML, which can either be provided
3292 in the XML or, if omitted, generated automatically.
3294 * qemu: Add virtiofs support
3296 This feature, introduced in QEMU 4.2, is a more modern alternative to
3297 virtio-9p, which is exposed through the same ``<filesystem/>`` element.
3299 * admin: Support reloading TLS certificates
3301 After renewing TLS certificates, it was usually necessary to restart
3302 libvirtd for the new ones to be loaded: now the same result can be obtained
3303 without restarting the daemon by using ``virt-admin server-update-tls`` .
3305 * **Removed features**
3307 * Removed support for INI style of comments
3309 With switching of our internal code to GLib, parsing of client
3310 authentication config files is handed over to GLib which does not support
3311 ``INI`` style of comments starting with a semicolon ( ``;`` ). Use number
3312 sign ( ``#`` ) instead.
3316 * qemu: Don't compare local and remote hostnames on migration
3318 This check was introduced to prevent same-host migration, but did not work
3319 as expected when multiple libvirtd instances were running on the same host
3320 but in different containers. With this release, the host UUID (which should
3321 be unique to the container) is checked instead.
3323 * qemu: Use per-VM event loops
3325 Instead of using a single even loop to process communication with the QEMU
3326 monitor and guest agent, create a separate one for each VM. This helps with
3327 scalability and prevents scenarios where a single malfunctioning VM could
3328 affect all those running on the same host.
3330 * qemu: Support migration with SLIRP helper interface
3332 With QEMU 5.0, a new D-Bus backend allows migration of external processes.
3333 When needed, libvirt will start a per-vm D-Bus bus, and migrate the
3334 slirp-helper along with QEMU.
3338 * qemu: Open backing chain late for shallow block copy reusing external
3341 With introduction of -blockdev for QEMU storage configuration in
3342 libvirt-5.10 we've started opening the backing chain of the
3343 destination/mirror of a virDomainBlockcopy started with
3344 VIR_DOMAIN_BLOCK_COPY_REUSE_EXT | VIR_DOMAIN_BLOCK_COPY_SHALLOW flags when
3345 starting the job rather than when virDomainBlockJobAbort with
3346 VIR_DOMAIN_BLOCK_JOB_ABORT_PIVOT is issued. For users depending on this
3347 undocumented quirky pre-blockdev behaviour this caused a regression as the
3348 backing chain could not be modified while the copy of the top image was
3349 progressing due to QEMU image locking. Note that this fix also requires
3350 qemu-5.0 while -blockdev is used starting from QEMU-4.2.
3352 * Don't generate machine names containing dots
3354 Even though the guest name containing dots is not a problem for libvirt
3355 itself, we need to strip them out when registering with machined because of
3356 the latter's requirements.
3364 * qemu: new rng backend type: builtin
3366 It implements qemu builtin rng backend. That uses getrandom syscall to
3367 generate random, no external rng source needed. Available since QEMU 4.2.
3369 * support for virtio+hostdev NIC <teaming>
3371 QEMU 4.2.0 and later, combined with a sufficiently recent guest virtio-net
3372 driver (e.g. the driver included in Linux kernel 4.18 and later), supports
3373 setting up a simple network bond device comprised of one virtio emulated
3374 NIC and one hostdev NIC (which must be an SRIOV VF). (in QEMU, this is
3375 known as the "virtio failover" feature). The allure of this setup is that
3376 the bond will always favor the hostdev device, providing better
3377 performance, until the guest is migrated - at that time QEMU will
3378 automatically unplug the hostdev NIC and the bond will send all traffic via
3379 the virtio NIC until migration is completed, then QEMU on the destination
3380 side will hotplug a new hostdev NIC and the bond will switch back to using
3381 the hostdev for network traffic. The result is that guests desiring the
3382 extra performance of a hostdev NIC are now migratable without network
3383 downtime (performance is just degraded during migration) and without
3384 requiring a complicated bonding configuration in the guest OS network
3385 config and complicated unplug/replug logic in the management application on
3386 the host - it can instead all be accomplished in libvirt with the interface
3387 <teaming> subelement "type" and "persistent" attributes.
3389 * support BR_ISOLATED flag for guest interfaces attached to a Linux host
3392 Since Linux kernel 4.18, the Linux host bridge has had a flag BR_ISOLATED
3393 that can be applied to individual ports. When this flag is set for a port,
3394 traffic is blocked between that port and any other port that also has the
3395 BR_ISOLATED flag set. libvirt domain interface config now supports setting
3396 this flag via the <port isolated='yes'/> setting. It can also be set for
3397 all connections to a particular libvirt network by setting the same option
3398 in the network config - since the port for the host itself does not have
3399 BR_ISOLATED set, the guests can communicate with the host and the outside
3400 world, but guests on that network can't communicate with each other. This
3401 feature works for QEMU and LXC guests with interfaces attached to a Linux
3404 * qemu: Introduce the 'armvtimer' timer type
3406 QEMU 5.0 introduces the ability to control the behavior of the virtual
3407 timer for KVM ARM/virt guests, and this new timer type exposes the same
3408 capability to libvirt users.
3410 * qemu: Storage configuration improvements
3412 Libvirt now accepts ``<backingStore type='volume'>`` and allows specifying
3413 the offset and size of the image format container inside the storage source
3414 via the ``<slices>`` subelement.
3416 * qemu: Introduce the 'tpm-spapr' TPM model
3418 This device, available starting from QEMU 5.0, is limited to pSeries
3421 * qemu: support Panic Crashloaded event handling
3423 The pvpanic device now supports a 'crashloaded' event, which is emitted
3424 when a guest panic has occurred but has already been handled by the guest
3427 * qemu: Implement virDomainGetHostnameFlags
3429 The ``--source`` argument to ``virsh domhostname`` can be used to specify
3430 what data source to use for the domain hostnames. Currently, in addition
3431 to the 'agent', libvirt can also use 'lease' information from dnsmasq to
3436 * qemu: Image format probing is allowed in certain cases
3438 To resolve regressions when users didn't specify the backing image format
3439 in the overlay, libvirt now probes the format in certain secure scenarios
3440 which fixes a few common existing cases. Additionally the knowledge base
3441 was extended to provide more information on how to rectify the problem.
3443 * qemu: Support "dies" in CPU topology
3445 This CPU topology concept, new in QEMU 4.1.0, sits between the existing
3446 "socket" and "core".
3448 * libxl: Add support for Credit2 scheduler parameters
3450 * lxc: Add support LXC 3 network configuration format
3454 * conf: Do not generate machine names ending with a dash
3456 Recent systemd versions do not allow them.
3458 * **Packaging changes**
3460 * use of gnulib has been completely eliminated
3462 Historically libvirt has embedded gnulib to provide fixes for various
3463 platform portability problems. This usage has now been eliminated and
3464 alternative approaches for platform portability problems adopted where
3465 required. This has been validated on the set of platforms covered by
3466 automated CI build testing. Other modern Linux distros using glibc are
3467 expected to work. Linux distros using non-glibc packages, and other
3468 non-Linux platforms may encounter regressions when building this release.
3469 Please report any build problems encountered back to the project
3470 maintainers for evaluation.
3476 * **Packaging changes**
3478 * support for python2 is removed
3480 Libvirt is no longer able to be built using the Python 2 binary. Python 3
3481 must be used instead.
3483 * docs: the python docutils toolset is now required
3485 The use of rst2html has been introduced for the website build process since
3486 docs are now being written in the RST as an alternative to HTML.
3490 * new PCI hostdev address type: unassigned
3492 A new PCI hostdev address type 'unassigned' is introduced. An unassigned
3493 PCI hostdev behaves like any regular PCI hostdev inside Libvirt, but it is
3494 not usable by the guest. This gives the user a new option to manage the
3495 binding of PCI devices via Libvirt, declaring PCI hostdevs in the domain
3496 XML but allowing just a subset of them to be assigned to the guest.
3498 * Provide init scripts for sub-deaemons
3500 So far libvirt shipped systemd unit files for sub-daemons. With this
3501 release, init scripts are available too. Package maintainers can choose
3502 which one to install via ``--with-init-script`` configure option.
3504 * qemu: Support cold-unplug of sound devices
3506 * qemu: Implement VIR_MIGRATE_PARAM_TLS_DESTINATION
3508 This flag, which can be enabled using ``virsh`` 's ``--tls-destination``
3509 option, allows migration to succeed in situations where there is a mismatch
3510 between the destination's hostname and the information stored in its TLS
3513 * qemu: Support reporting memory bandwidth usage stats
3515 Implement Intel RDT-MBM in libvirt. The stats can be obtained via ``virsh
3516 domstats --memory`` .
3518 * qemu: Allow accessing NVMe disks directly
3520 Before this release there were two ways to configure a NVMe disk for a
3521 domain. The first was using <disk/> with the <source/> pointing to the
3522 ``/dev/nvmeXXXX`` . The other was using PCI assignment via <hostdev/>
3523 element. Both have their disadvantages: the former adds latency of file
3524 system and block layers of the host kernel, the latter prohibits domain
3525 migration. In this release the third way of configuring NVMe disk is added
3526 which combines the advantages and drops disadvantages of the previous two
3527 ways. It's accessible via <disk type='nvme'/>.
3529 * **Removed features**
3531 * 'phyp' Power Hypervisor driver removed
3533 The 'phyp' Power Hypervisor driver has not seen active development since
3534 2011 and does not seem to have any real world usage. It has now been
3539 * qemu: xz save image compression is faster
3541 When using the xz format to compressed virtual machine saved state images,
3542 the "-3" compression level preset is now used. This results in slightly
3543 larger files, but with a massively reduced time to compress. The xz format
3544 offers the best compression level for saved state images, albeit still with
3545 the slowest running time. For the fastest possible running time, at cost of
3546 the larest compressed size, lzop should be used.
3548 * domain: Improve job stat handling
3550 It is now possible to retrieve stats for completed and failed jobs.
3552 * qemu: Don't hold monitor and agent job at the same time
3554 Before this change, a malicious (or buggy) ``qemu-guest-agent`` running in
3555 the guest could make other libvirt APIs unavailable for an unbounded amount
3560 * qemu: Report error if backing image format is not specified explicitly
3562 For a long time libvirt was assuming that a backing file is RAW when the
3563 format was not specified. This didn't pose a problem until blockdev support
3564 was enabled in last release. Libvirt now requires that the format is
3565 specified in the image metadata or domain XML and the VM will refuse to
3566 start otherwise. Additionally the error message now links to the knowledge
3567 base which summarizes how to fix the images.
3569 * qemu: Fix non-shared storage migration over NBD
3571 * qemu: Generate a single MAC address for hotplugged network devices
3573 Since libvirt 4.6.0, when hotplugging a network device that didn't have a
3574 MAC address already assigned by the user, two separate addresses would be
3575 generated: one for the live configuration, which would show up immediately,
3576 and one for the inactive configuration, which would show up after the first
3577 reboot. This situation was clearly undesirable, so a single MAC address is
3578 now generated and used both for the live configuration and the inactive
3582 v5.10.0 (2019-12-02)
3583 ====================
3587 * qemu: Introduce support for ARM CPU features
3589 The only features supported at the moment are SVE vector lengths, which
3590 were introduced in QEMU 4.2.0.
3592 * qemu: Support boot display for GPU mediated devices
3594 Until now, GPU mediated devices generally did not show any output until the
3595 guest OS had initialized the vGPU. By specifying the ``ramfb`` attribute,
3596 QEMU can be configured to use ramfb as a boot display for the device: this
3597 allows for display of firmware messages, boot loader menu, and other output
3598 before the guest OS has initialized the vGPU.
3600 * Add API to change the response timeout for guest agent commands
3602 By default, when a command is sent to the guest agent, libvirt waits
3603 forever for a response from the guest agent. If the guest is unresponsive
3604 for any reason, this can block the calling thread indefinitely. By setting
3605 a custom timeout using ``virDomainAgentSetResponseTimeout()`` , API users
3606 can change this behavior.
3610 * Devices CGroup v2 support
3612 Libvirt supported all controllers of CGroup v2 but the devices controller
3613 which is implemented in this release.
3615 * Cold plug of sound device
3617 The QEMU driver now can handle cold plug of ``<sound/>`` devices.
3619 * Probe for default CPU types
3621 With QEMU 4.2.0 we can probe for the default CPU model used by QEMU for a
3622 particular machine type and store it in the domain XML. This way the chosen
3623 CPU model is more visible to users and libvirt will make sure the guest
3624 will see the exact same CPU after migration.
3626 * Adaptation to qemu's blockdev
3628 QEMU introduced a new way of specifying disks on the command line which
3629 enables fine-grained control over the block stack. Libvirt has adapted to
3634 * More GLib integration
3636 More patches were merged that replace our internal functions with GLib
3637 ones. Also some effort was invested in replacing gnulib modules with GLib
3640 * Rewrite of Perl scripts into Python
3642 Libvirt used Perl scripts to check for coding style, generate some code and
3643 things like that. To bring the number of languages used down, these scripts
3644 were rewritten into Python.
3648 * Warn verbosely if using old loader:nvram pairs
3650 Some distributions still use ``--with-loader-nvram`` or ``nvram`` variable
3651 in qemu.conf. This is now discouraged in favour of FW descriptors. However,
3652 instead of silently ignoring user's config, libvirt warns if outdated
3655 * Drop pconfig from Icelake-Server CPU model
3657 The pconfig feature was enabled in QEMU by accident in 3.1.0. All other
3658 newer versions do not support it and it was removed from the Icelake-Server
3661 * Wait longer for device removal confirmation on PPC64
3663 After sending device hot unplug request to QEMU, libvirt waits up to 5
3664 seconds for qemu to confirm the device removal. On some architectures (like
3665 PPC64) this can take longer time and libvirt now reflects that.
3667 * Forcibly create nodes in domain's namespace
3669 The QEMU driver starts a domain in a namespace with private ``/dev`` and
3670 creates only those nodes there which the domain is configured to have.
3671 However, it may have happened that if a node changed its minor number this
3672 change wasn't propagated to the namespace.
3674 * Various AppArmor bugfixes
3676 The AppArmor driver now knows how to handle ``<shmem/>`` devices and also
3677 snapshotting more disks at once.
3679 * Improved video model autoselection
3681 If a graphics device was added to XML that had no video device, libvirt
3682 automatically added a video device which was always of type 'cirrus' on
3683 x86_64, even if the underlying qemu didn't support cirrus. Libvirt now
3684 bases the decision on qemu's capabilities.
3690 * **Packaging changes**
3692 * Start linking against GLib and using its features
3694 Up until now, libvirt has been dealing with platform portability and the
3695 lack of certain features in libc by using gnulib and implementing its own
3696 functions and data structures respectively; going forward, it will prefer
3697 the facilities offered by GLib instead.
3699 * Stop distributing generated documentation
3701 Most downstreams already patch the libvirt source to some extent, so this
3702 change will probably not affect them.
3704 * Rewrite several Perl scripts in Python
3706 Phasing out Perl usage is part of the project strategy.
3710 * qemu: Introduce a new video model of type 'ramfb'
3712 Introduce a new video model type to the domain XML that supports the
3713 ``ramfb`` standalone device in qemu.
3715 * qemu: Implement the ccf-assist pSeries feature
3717 Users can now decide whether ccf-assist (Count Cache Flush Assist) support
3718 should be available to pSeries guests.
3720 * Xen: Support specifying ACPI firmware path
3722 The libxl driver now supports specifying an ACPI firmware path using the
3725 * qemu: Support specifying resolution for video devices
3727 * **Removed features**
3729 * logging: Drop support for including stack traces
3731 This feature was intended to aid debugging, but in practice it resulted in
3732 logs that were too verbose to be useful and also resulted in a significant
3733 performance penalty.
3737 * qemu: Implement CPU comparison/baseline on s390x
3739 This functionality has been historically limited to x86_64, but it's now
3740 available on s390x too.
3744 * lib: autostart objects exactly once
3746 If libvirtd or any of the sub-daemons is started with socket activation
3747 then objects might be autostarted more than once. For instance, if a domain
3748 under ``qemu:///session`` URI is mark as autostarted and the session daemon
3749 is started then the domain is started with it. If user shuts the domain
3750 down and the session daemon is started again, the user's wish to keep the
3751 domain shut off is ignored and the domain is autostarted again. This is now
3754 * qemu: Properly advertise bochs-display availability
3756 Support for ``bochs-display`` was introduced in libvirt 5.6.0, but until
3757 now the model was not listed in the domain capabilities.
3759 * security: Don't remember labels for TPM devices
3761 Due to the way they're implemented in the kernel, trying to remember labels
3762 for TPM devices makes it impossible to use them.
3764 * security: Properly rollback after failure in a stacked driver
3766 When multiple security drivers are involved, failure in one of them would
3767 result in only the corresponding changes being rolled back, leaving the
3768 ones performed by drivers that had been activated earlier in place. All
3769 changes are rolled back now.
3771 * Fix build with musl libc
3773 * Improve compatibility with non-bash shells
3781 * qemu: Support use of precreated tap/macvtap devices by unprivileged
3784 It is now possible for an unprivileged libvirtd to make use of tap and
3785 macvtap devices that were previously created by some other entity. This is
3786 done by setting ``managed='no'`` along with the device name in the
3787 ``target`` subelement of ``<interface type='ethernet'>`` .
3789 * qemu: Support vhost-user-gpu
3791 Support for running virtio GPUs in separate processes with vhost-user
3792 backend. It requires QEMU newer than 4.1.
3794 * Introduce virConnectSetIdentity API
3796 When split daemons are in use, this API is used to forward uid, gid and
3797 SELinux info from ``virproxyd`` to other driver daemons such as
3802 * qemu: Support running SLIRP networking in a separate process
3804 User can configure the slirp-helper path in ``qemu.conf`` . It will start a
3805 slirp-helper process to provide SLIRP networking when the VM is started
3806 with network interface "user". That will allow stricter security policies
3807 for QEMU SLIRP network.
3809 * **Removed features**
3811 * Remove xenapi driver
3813 The xenapi driver is removed since it has not received any significant
3814 development since its initial contribution nine years ago and has no known
3823 * qemu: Support Direct Mode for Hyper-V Synthetic timers
3825 The QEMU driver now supports Direct Mode for Hyper-V Synthetic timers for
3828 * lib: Add virDomainGetGuestInfo()
3830 This API is intended to aggregate several guest agent information queries
3831 and is inspired by stats API ``virDomainListGetStats()`` . It is
3832 anticipated that this information will be provided by a guest agent running
3833 within the domain. It's exposed as ``virsh guestinfo`` .
3835 * Experimental split of libvirtd into separate daemons
3837 The big monolithic libvirtd daemon can now be replaced by smaller
3838 per-driver daemons. The new split daemons are considered experimental at
3839 this time and distributions are encouraged to continue using the
3840 traditional libvirtd by default.
3842 * qemu: Support kvm-hint-dedicated performance hint
3844 With ``<hint-dedicated state='on'/>`` and ``<cpu
3845 mode='host-passthrough'/>`` , it allows a guest to enable optimizations
3846 when running on dedicated vCPUs. QEMU newer than 2.12.0 and kernel newer
3847 than 4.17 are required.
3849 * **Removed features**
3851 * Remove KVM assignment support
3853 The KVM style of PCI device assignment was removed from the kernel in
3854 version 4.12.0 after being deprecated since 4.2.0. Libvirt defaults to VFIO
3855 for a long time. Remove support for KVM device assignment from libvirt too.
3857 * libxml: min required libxml is now 2.9.1
3859 Support for building with libxml versions older than 2.9.1 has been
3864 * virsh: Support setting bandwidth in migrate subcommand
3866 In addition to postcopy bandwidth, the ``virsh migrate`` subcommand now
3867 supports specifying precopy bandwidth with the ``--bandwidth`` parameter.
3869 * libxl: Implement domain metadata getter/setter
3871 The libxl driver now supports ``virDomainGetMetadata()`` and
3872 ``virDomainSetMetadata()`` APIs.
3874 * test driver: Expand API coverage
3876 Additional APIs have been implemented in the test driver.
3878 * Report RNG device in domain capabilities XML
3880 Libvirt now reports if RNG devices are supported by the underlying
3881 hypervisor in the domain capabilities XML.
3883 * Stop linking NSS plugins with libvirt.so
3885 This reduces the amount of code and 3rd party libraries are that loaded
3888 * Split the setuid virt-login-shell binary into two pieces
3890 The setuid virt-login-shell binary is now a tiny shim that sanitizes the
3891 process execution environment variables and arguments, before launching the
3892 trusted virt-login-shell-helper binary.
3894 * qemu: Allow migration with disk cache on
3896 When QEMU supports flushing caches at the end of migration, we can safely
3897 allow migration even if ``disk/driver/@cache`` is neither ``none`` nor
3902 * Various security label remembering fixes
3904 In the previous release libvirt introduced remembering of original owners
3905 and SELinux labels on files. However, the feature did not work properly
3906 with snapshots, on migrations or on network filesystems. This is now fixed.
3908 * Allow greater PCI domain numbers
3910 Libvirt used to require PCI domain number to be not greater than 0xFFFF.
3911 The code was changed to allow 32 bits long numbers.
3913 * Various D-Bus fixes
3915 When D-Bus is not available, libvirt was reporting random errors. These are
3918 * Prefer read-only opening of PCI config files
3920 When enumerating PCI bus, libvirt opens config files under ``sysfs`` mount
3921 and parses them to learn various aspects of the device (e.g. its
3922 capabilities). Only in a very limited number of cases it is actually
3923 writing into the file. However, it used to open the file also for writing
3924 even if it was only reading from it.
3926 * Fix AppArmor profile
3928 Since the ``5.6.0`` release, libvirt uses ``procfs`` to learn the list of
3929 opened file descriptors when spawning a command. However, our AppArmor
3930 profile was not allowing such access.
3932 * Don't block storage driver when starting or building a pool
3934 Starting or building a storage pool can take a long time to finish. During
3935 this time the storage driver was blocked and thus no other API involving
3936 the storage driver could run. This is now fixed.
3944 * qemu: Introduce a new video model of type 'bochs'
3946 Introduce a new video model type that supports the ``bochs-display`` device
3947 that was added in qemu version 3.0.
3949 * api: new virDomainCheckpoint APIs
3951 Introduce several new APIs for creating and managing checkpoints in the
3952 test and qemu drivers (the latter requires qcow2 images). Checkpoints serve
3953 as a way to tell which portions of a disk have changed since a point in
3956 * qemu: Add support for overriding max threads per process limit
3958 systemd-based systems impose a limit on the number of threads a process can
3959 spawn, which in some cases can be exceeded by QEMU processes running VMs.
3960 Add a ``max_threads_per_process`` option to qemu.conf to override the
3963 * Remember original owners and SELinux labels of files
3965 When a domain is starting up libvirt changes DAC and SELinux labels so that
3966 domain can access it. However, it never remembered the original labels and
3967 therefore the file was returned back to ``root:root`` . With this release,
3968 the original labels are remembered and restored properly.
3970 * network: Allow passing arbitrary options to dnsmasq
3972 This works similarly to the existing support for passing arbitrary options
3973 to QEMU, and just like that feature it comes with no support guarantees.
3975 * **Removed features**
3977 * xen: Remove sxpr config support
3979 Remove the sxpr style config parser and formatter a year after the xend
3984 * qemu: Allow XML validation for snapshot creation
3986 Add flag ``VIR_DOMAIN_SNAPSHOT_CREATE_VALIDATE`` to validate snapshot input
3987 XML. For virsh, users can use it as ``virsh snapshot-create --validate`` .
3989 * Support encrypted soft TPM
3991 A soft TPM backend could be encrypted with passphrase. Now libvirt supports
3992 using a ``secret`` object to hold the passphrase, and referring to it via
3993 the ``encryption`` element of the TPM device.
3995 * test driver: Expand API coverage
3997 Additional APIs have been implemented in the test driver.
3999 * Implement per-driver locking
4001 Drivers now acquire a lock when they're loaded, ensuring that there can
4002 never be two instances of the same driver active at a time.
4004 * nss: Report newer addresses first
4006 In some cases, a guest might be assigned a new IP address by DHCP before
4007 the previous lease has expired, in which case the NSS plugin will correctly
4008 report both addresses; many applications, however, ignore all addresses but
4009 the first, and may thus end up trying to connect using a stale address. To
4010 prevent that from happening, the NSS plugin will now always report the
4011 newest address first.
4013 * util: Optimize mass closing of FDs when spawning child processes
4015 When the limit on the number of FDs is very high, closing all unwanted FDs
4016 after calling ``fork()`` can take a lot of time and delay the start of the
4017 child process. libvirt will now use an optimized algorithm that minimizes
4022 * logging: Ensure virtlogd rollover takes priority over logrotate
4024 virtlogd implements its own rollover mechanism, but until now logrotate
4025 could end up acting on the logs before virtlogd had a chance to do so
4034 * api: Prevent access to several APIs over read-only connections
4036 Certain APIs give root-equivalent access to the host, and as such should be
4037 limited to privileged users. CVE-2019-10161, CVE-2019-10166,
4038 CVE-2019-10167, CVE-2019-10168.
4042 * qemu: Support SMMUv3 IOMMU
4044 SMMUv3 is an IOMMU implementation for ARM virt guests.
4046 * network: Introduce the network port API
4048 This new public API can be used by virtualization drivers to manage network
4049 resources associated with guests, and is a further step towards splitting
4050 libvirtd into multiple daemons.
4052 * **Removed features**
4054 * qemu: Remove support for virDomainQemuAttach and
4055 virConnectDomainXMLFromNative APIs
4057 The qemu implementations for the APIs mentioned above were removed and the
4058 APIs now return an error. The implementation was stale for a long time and
4059 did not work with modern QEMU command lines, generated from libvirt or
4062 * Stop supporting migration of config files from pre-XDG layout
4064 The new layout was introduced with libvirt 0.9.13 (Jul 2012).
4066 * Remove Avahi mDNS support
4068 This feature was never used outside of virt-manager, which has itself
4069 stopped using it a while ago.
4073 * sysinfo: Report SMBIOS information on aarch64
4075 While SMBIOS support has historically been limited to x86_64, modern
4076 aarch64 machines often offer access to the same information as well, and
4077 libvirt now exposes it to the user when that's the case.
4079 * test driver: Expand API coverage
4081 Even more APIs that were missing from the test driver have now been
4084 * virt-xml-validate: Allow input to be read from stdin
4086 * qemu: Validate spapr-vio addresses as 32-bit
4088 libvirt has always considered these addresses (used for pSeries guests) as
4089 64-bit, but the sPAPR specification says that they're 32-bit instead.
4093 * qemu: Set process affinity correctly when using <numatune>
4095 libvirt would mistakenly interpret the ``nodeset`` attribute as a list of
4096 CPUs instead of as a list of NUMA node, and the process affinity would be
4097 set incorrectly as a result; this has now been fixed.
4105 * cpu: Introduce support for the md-clear CPUID bit
4107 This bit is set when microcode provides the mechanism to invoke a flush of
4108 various exploitable CPU buffers by invoking the x86 ``VERW`` instruction.
4109 CVE-2018-12126, CVE-2018-12127, CVE-2018-12130, CVE-2019-11091.
4111 * Restrict user access to virt-admin, virtlogd and virtlockd
4113 The intended users for these facilities are the ``root`` user and the
4114 ``libvirtd`` service respectively, but these restrictions were not enforced
4115 correctly. CVE-2019-10132.
4119 * test driver: Expand API coverage
4121 Several APIs that were missing from the test driver have now been
4124 * Avoid unnecessary static linking
4126 Most binaries shipped as part of libvirt, for example ``virtlogd`` and
4127 ``libvirt_iohelper`` , were embedding parts of the library even though they
4128 also linked against the ``libvirt.so`` dynamic library. This is no longer
4129 the case, which results in both the disk and memory footprint being
4132 * qemu: Report stat-htlb-pgalloc and stat-htlb-pgfail balloon stats
4134 These stats have been introduced in QEMU 3.0.
4138 * qemu: Fix emulator scheduler support
4140 Setting the scheduler for QEMU's main thread before QEMU had a chance to
4141 start up other threads was misleading as it would affect other threads
4142 (vCPU and I/O) as well. In some particular situations this could also lead
4143 to an error when the thread for vCPU #0 was being moved to its cpu,cpuacct
4144 cgroup. This was fixed so that the scheduler for the main thread is set
4147 * apparmor: Allow hotplug of vhost-scsi devices
4155 * qemu: Add support for setting the emulator scheduler parameters
4157 I/O threads and vCPU threads already support setting schedulers, but until
4158 now it was impossible to do so for the main QEMU thread (emulator thread in
4159 the libvirt naming). This is, however, requested for some very specific
4160 scenarios, for example when vCPU threads are running at such priority that
4161 could starve the main thread.
4163 * **Removed features**
4165 * vbox: Drop support for VirtualBox 4.x releases
4167 Support for all the 4.x releases was ended by VirtualBox maintainers in
4168 December 2015. Therefore, libvirt support for these releases is dropped.
4172 * qemu: Use PCI by default for RISC-V guests
4174 PCI support for RISC-V guests was already available in libvirt 5.1.0, but
4175 it required the user to opt-in by manually assigning PCI addresses: with
4176 this release, RISC-V guests will use PCI automatically when running against
4177 a recent enough (4.0.0+) QEMU release.
4179 * qemu: Advertise firmware autoselection in domain capabilities
4181 The firmware autoselection feature is now exposed in domain capabilities
4182 and management applications can query for accepted values, i.e. values that
4183 are accepted and for which libvirt found firmware descriptor files.
4184 Firmware Secure Boot support is also advertised.
4186 * Drop YAJL 1 support
4188 YAJL 2 is widely adopted and maintaining side by side support for two
4189 versions is unnecessary.
4193 * rpc: cleanup in virNetTLSContextNew
4195 Failed new gnutls context allocations in virNetTLSContextNew function
4196 results in double free and segfault. Occasional memory leaks may also
4199 * virsh: various completers fixes
4201 There were some possible crashers, memory leaks, etc. which are now fixed.
4203 * qemu: Make hugepages work with memfd backend
4205 Due to a bug in command line generation libvirt did not honor hugepages
4206 setting with memfd backend.
4208 * Enforce ACL write permission for getting guest time & hostname
4210 Getting the guest time and hostname both require use of guest agent
4211 commands. These must not be allowed for read-only users, so the permissions
4212 check must validate "write" permission not "read".
4220 * Add Storage Pool Capabilities output
4222 Add support to list an enumerated list of supported Storage Pools via the
4223 virConnectGetCapabilities API when connected via a Storage Driver. Add
4224 support to get a more detailed list XML output Storage Pool Capabilities
4225 vis the virConnectGetStoragePoolCapabilites API.
4227 * qemu: Support virtio-{non-}transitional device models
4229 ``virtio-transitional`` and ``virtio-non-transitional`` ``model`` values
4230 were added to the QEMU driver for the following devices: ``disk`` ,
4231 ``interface`` , ``filesystem`` , ``rng`` , ``vsock`` , ``memballoon`` ,
4232 ``controller`` type ``scsi`` , ``controller`` type ``virtio-serial`` ,
4233 ``input`` bus ``virtio`` type ``passthrough`` , ``hostdev`` type
4234 ``scsi_host`` . These new models can be used to give fine grained control
4235 over what virtio device version is presented to the guest.
4237 * qemu: Enable firmware autoselection
4239 Libvirt allows users to provide loader path for some time now. However,
4240 this puts some burden on users because they need to know what firmware
4241 meets their requirements. Now that QEMU ships firmware description files
4242 this burden can be moved onto libvirt. It is as easy as setting the
4243 ``firmware`` attribute in the ``os`` element (accepted values are ``bios``
4244 and ``efi`` ). Moreover, libvirt automatically enables domain features
4245 needed for firmware it chooses.
4247 * snapshots: Add support for topological listings
4249 A new flag VIR_DOMAIN_SNAPSHOT_LIST_TOPOLOGICAL is available for the
4250 various snapshot listing APIs such as virDomainListAllSnapshots(). For
4251 drivers that support the flag, the listed snapshots are guaranteed to be
4252 sorted such that parents occur before children.
4254 * Xen: Add support for max grant frames setting
4256 Add support for Xen's max_grant_frames setting by adding a new xenbus
4257 controller type with a maxGrantFrames attribute. E.g. ``<controller
4258 type='xenbus' maxGrantFrames='64'/>``
4260 * qemu: Add support for parallel migration
4262 With QEMU 4.0.0 libvirt can enable parallel migration which causes the
4263 memory pages to be processed in parallel by several threads and sent to the
4264 destination host using several connections at the same time. This may
4265 increase migration speed in case a single thread is unable to saturate the
4268 * **Removed features**
4270 * Drop support for Upstart and "Red Hat" init scripts
4272 Not a single one of the platforms we target still uses Upstart, and the
4273 Upstart project itself has been abandoned for several years now; the same
4274 is true for the "Red Hat" (really System V) init scripts, since RHEL 7 and
4275 later releases use systemd.
4279 * Report class information for PCI node device capability.
4281 * Split setup of IPv4 and IPv6 top level chain
4283 The requirement resulting from private chains improvement done in
4284 ``v5.1.0`` was refined so that only tables from corresponding IP version
4285 are required. This means that if a network doesn't have ``IPv6`` enabled
4286 then those tables are not required.
4288 * Don't default to building the QEMU driver
4290 Historically, the QEMU driver has been special in that it was enabled by
4291 default, with the option to explicitly opt-out of it; starting now, we're
4292 enabling it opportunistically if we detect that all requirements are
4293 available, just like we do with other drivers.
4297 * virt-host-validate: Fix IOMMU check on s390x
4299 * qemu: Allow creating pSeries guests with graphics and no USB mouse
4301 It's now possible to prevent libvirt from automatically adding a USB mouse
4302 to pSeries guests by including a USB tablet in the input XML: doing so is
4303 desiderable as using a tablet results in a much better user experience when
4306 * qemu: Set $HOME and XGD variables for qemu:///system guests
4308 This avoids files being accidentally created under ``/`` or the guests not
4309 being able to start because they lack the necessary permissions to write to
4318 * bhyve: Add support for additional command-line arguments
4320 The bhyve driver now supports passing additional command-line arguments to
4321 the bhyve process using the new ``<bhyve:commandline>`` element in domain
4324 * network: Support setting a firewalld "zone" for virtual network bridges
4326 All libvirt virtual networks with bridges managed by libvirt (i.e. those
4327 with forward mode of "nat", "route", "open", or no forward mode) will now
4328 be placed in a special firewalld zone called "libvirt" by default. The zone
4329 of any network bridge can be changed using the ``zone`` attribute of the
4330 network's ``bridge`` element.
4332 * bhyve: Support for ignoring unknown MSRs reads and writes
4334 A new <features> element <msrs unknown='ignore'/> was introduced and the
4335 bhyve driver supports it to control unknown Model Specific Registers (MSRs)
4338 * qemu: Add support for encrypted VNC TLS keys
4340 Use the password stored in the secret driver under the uuid specified by
4341 the ``vnc_tls_x509_secret_uuid`` option in qemu.conf.
4343 * Add storage pool namespace options
4345 Allow for adjustment of RBD configuration options via Storage Pool XML
4346 Namespace adjustments.
4348 * qemu: Add support for setting post-copy migration bandwidth
4350 Users can now limit the bandwidth of post-copy migration, e.g. via ``virsh
4351 migrate --postcopy-bandwidth`` .
4355 * Create private chains for virtual network firewall rules
4357 Historically firewall rules for virtual networks were added straight into
4358 the base chains. This works but has a number of bugs and design
4359 limitations. To address them, libvirt now puts firewall rules into its own
4360 chains. Note that with this change the ``filter`` , ``nat`` and ``mangle``
4361 tables are required for both ``IPv4`` and ``IPv6`` .
4363 * Detect CEPH and GPFS as shared FS
4365 When starting a migration libvirt performs some sanity checks to make sure
4366 domain will be able to run on the destination. One of the requirements is
4367 that the disk has to either be migrated too or be accessible from a network
4368 filesystem. CEPH and GPFS weren't detected as a network filesystem.
4370 * Advertise network MTU via DHCP when specified
4372 If network MTU is set and the network has DHCP enabled, advertise the MTU
4373 in DHCP transaction too so that clients can adjust their link accordingly.
4375 * qemu: Allocate memory at the configured NUMA nodes from start
4377 Libvirt used to just start QEMU, let it allocate memory for the guest, and
4378 then use CGroups to move the memory to configured NUMA nodes. This is
4379 suboptimal as huge chunks of memory have to be moved. Moreover, this relies
4380 on ability to move memory later which is not always true. A change was made
4381 to set process affinity correctly from the start so that memory is
4382 allocated on the configured nodes from the beginning.
4384 * Support for newer Wireshark
4386 Adapt libvirt to use the more recent release requiring a source build
4387 configuration of libvirt ``--with-wireshark`` to upgrade to the more recent
4390 * Batch mode virsh and virt-admin parsing improvements
4392 When parsing a single-argument command_string in batch mode, virsh and
4393 virt-admin now permit newlines in addition to semicolons for splitting
4394 commands, and backslash-newline for splitting long lines, to be more like
4399 * qemu: Use CAP_DAC_OVERRIDE during QEMU capabilities probing
4401 By default, libvirt runs the QEMU process as ``qemu:qemu`` which could
4402 cause issues during probing as some features like AMD SEV might be
4403 inaccessible to QEMU because of file system permissions. Therefore,
4404 ``CAP_DAC_OVERRIDE`` is granted to overcome these for the purposes of
4407 * storage: Add default mount options for fs/netfs storage pools
4409 Altered the command line generation for fs/netfs storage pools to add some
4410 default options. For Linux based systems, the options added are "nodev,
4411 nosuid, noexec". For FreeBSD based systems, the options added are "nosuid,
4414 * qemu: Allow use of PCI for RISC-V guests
4416 This works with QEMU 4.0.0+ only and is opt-in at the moment, since it
4417 requires users to manually assign PCI addresses, but is otherwise fully
4420 * network: Fix virtual networks on systems using firewalld+nftables
4422 Because of the transitional state of firewalld's new support for nftables,
4423 not all iptables features required by libvirt are yet available, so libvirt
4424 must continue to use iptables for its own packet filtering rules even when
4425 the firewalld backend is set to use nftables. However, due to the way
4426 iptables support is implemented in kernels using nftables (iptables rules
4427 are converted to nftables rules and processed in a separate hook from the
4428 native nftables rules), guest networking was broken on hosts with firewalld
4429 configured to use nftables as the backend. This has been fixed by putting
4430 libvirt-managed bridges in their own firewalld zone, so that guest traffic
4431 can be forwarded beyond the host and host services can be exposed to guests
4432 on the virtual network without opening up those same services to the rest
4433 of the physical network. This means that host access from virtual machines
4434 is no longer controlled by the firewalld default zone (usually "public"),
4435 but rather by the new firewalld zone called "libvirt" (unless configured
4436 otherwise using the new zone attribute of the network bridge element).
4438 * qemu: Fix i6300esb watchdog hotplug on Q35
4440 Ensure that libvirt allocates a PCI address for the device so that QEMU did
4441 not default to an address that would not allow for device hotplug.
4443 * lxc: Don't reboot host on virDomainReboot
4445 If the container is really a simple one (init is just bash and the whole
4446 root is passed through) then virDomainReboot and virDomainShutdown would
4447 reboot or shutdown the host. The solution is to use different method to
4448 reboot or shutdown the container in that case (e.g. signal).
4450 * rpc: Various stream fixes
4452 One particular race was fixed, one locking problem and error reporting from
4453 streams was made better.
4455 * qemu: Fix guestfwd hotplug/hotunplug
4457 Fixed the generation of the guestfwd hotplug/unplug command sent to QEMU to
4458 match the syntax used when creating the initial command line.
4460 * qemu: Forbid CDROMs on virtio bus
4462 Attempting to create an empty virtio-blk drive or attempting to eject it
4463 results into an error. Forbid configurations where users would attempt to
4464 use CDROMs in virtio bus.
4466 * qemu: Use 'raw' for 'volume' disks without format
4468 Storage pools might want to specify format of the image when translating
4469 the volume thus libvirt can't add any default format when parsing the XML.
4470 Add an explicit format when starting the VM and format is not present
4471 neither by user specifying it nor by the storage pool translation function.
4473 * qemu: Assume 'raw' default storage format also for network storage
4475 Post parse callback adds the 'raw' type only for local files. Remote files
4476 can also have backing store (even local) so we should do this also for
4477 network backed storage.
4479 * qemu: Fix block job progress reporting and advocate for READY event
4481 In some cases QEMU can get to 100% and still not reach the synchronised
4482 phase. Initiating a pivot in that case will fail. Therefore it is strongly
4483 advised to wait for ``VIR_DOMAIN_BLOCK_JOB_READY`` event which does not
4484 suffer from this problem.
4486 * qemu: Don't format image properties for empty drive
4488 If a ``-drive`` has no image, then formatting attributes such as cache,
4489 readonly, etc. would cause errors to be reported from QEMU. This was fixed
4490 by not supplying the attributes for devices without an image.
4492 * External snapshot metadata redefinition is fixed
4494 Attempting to use VIR_DOMAIN_SNAPSHOT_CREATE_REDEFINE to reinstate the
4495 metadata describing an external snapshot created earlier for an offline
4496 domain no longer fails.
4504 * Xen: Add support for openvswitch
4506 The libxl driver now supports virtual interfaces that connect to an
4507 openvswitch bridge, including interfaces with VLAN tagging and trunking
4510 * qemu: Report whether KVM nesting is available
4512 Running nested KVM guests requires specific configuration steps to be
4513 performed on the host; libvirt will now report in the host capabilities
4514 whether KVM nesting support is available.
4516 * **Removed features**
4520 The UML driver was unmaintained and not tested for quite some time now.
4521 Worse, there is a bug that causes it to deadlock on some very basic
4522 operations (e.g. dumping domain XML). These facts make us believe no one
4527 * qemu: Add support for ARMv6l guests
4529 * Support more NVDIMM configuration options
4531 Introduce more configuration options. For the source element, add the
4532 'alignsize' and 'pmem' subelements. For the target element, add the
4533 'readonly' subelement.
4535 * cpu: Add support for "stibp" x86_64 feature
4537 Add cpu flag stibp (Single Thread Indirect Branch Predictors) to prevent
4538 indirect branch predictions from being controlled by the sibling
4541 * libxl: Handle external domain destroy
4543 Historically, if a domain was destroyed using ``xl`` rather than through
4544 libvirt APIs, libvirt would not be aware of the fact and keep considering
4545 it as running. This is no longer the case.
4547 * Start selecting the first available DRI device for OpenGL operations
4549 If OpenGL support is needed (either with SPICE gl enabled or with
4550 egl-headless), libvirt is now able to pick the first available DRI device
4551 for the job. At the same time, this improvement is also a bugfix as it
4552 prevents permission-related issues with regards to our mount namespaces and
4553 the default DRI render node's permissions which would normally prevent QEMU
4554 from accessing such a device.
4556 * qemu: Add support for postcopy-requests migration statistics
4558 The ``virDomainJobInfo`` can get number page requests received from the
4559 destination host during post-copy migration.
4563 * lxc: Don't forbid interfaces with type=direct
4565 Such interfaces are supported by lxc and should be allowed.
4567 * qemu: Fully clean up RNG devices on detach
4569 Some RNG device types, such as those using EGD, might need extra clean up
4570 on the host in addition to removing the guest-side device.
4573 v4.10.0 (2018-12-03)
4574 ====================
4578 * qemu: Add Hyper-V PV IPI and Enlightened VMCS support
4580 The QEMU driver now has support for Hyper-V PV IPI and Enlightened VMCS for
4581 Windows and Hyper-V guests.
4583 * qemu: Added support for PCI devices on S390
4585 PCI addresses can now include the new zpci element which contains uid
4586 (user-defined identifier) and fid (PCI function identifier) attributes and
4587 makes the corresponding devices usable by S390 guests.
4589 * Support changing IOThread polling parameters for a live guest
4591 Introduced virDomainSetIOThreadParams which allows dynamically setting the
4592 IOThread polling parameters used by QEMU to manage the thread polling
4593 interval and the algorithm for growth or shrink of the polling time. The
4594 values only affect a running guest with IOThreads. The guest's IOThread
4595 polling values can be viewed via the domain statistics.
4597 * Xen: Add support for PVH
4599 The libxl driver now supports Xen's PVH virtual machine type. PVH machines
4600 are enabled with the new "xenpvh" OS type, e.g.
4601 ``<os><type>xenpvh</type></os>``
4603 * qemu: Added support for CMT (Cache Monitoring Technology)
4605 Introduced cache monitoring using the ``monitor`` element in ``cachetune``
4606 for vCPU threads. Added interfaces to get and display the cache utilization
4607 statistics through the command 'virsh domstats' via the
4608 virConnectGetAllDomainStats API.
4610 * qemu: Add support for nested HV for pSeries guests
4612 Nested HV support makes it possible to run nested (L2) guests with minimal
4613 performance penalty when compared to regular (L1) guests on ppc64 hardware.
4617 * Xen: Handle soft reset shutdown event
4619 The pvops Linux kernel uses soft reset to handle the crash machine
4620 operation. The libxl driver now supports the soft reset shutdown event,
4621 allowing proper crash handling of pvops-based HVM domains.
4629 * util: Add cgroup v2 support
4631 cgroup v2 support has been implemented in libvirt, with both "unified" (v2
4632 only) and "hybrid" (v2 + v1) setups being usable; existing "legacy" (v1
4633 only) setups will keep working.
4635 * qemu: Add vfio AP support
4637 The QEMU driver now has support to passthrough adjunct processors into QEMU
4642 * rpc: Make 'genprotocol' output reproducible
4644 This is another step towards making libvirt builds fully reproducible.
4648 * security: Fix permissions for UNIX sockets
4650 Since 4.5.0, libvirt is using FD passing to hand sockets over to QEMU,
4651 which in theory removes the need for them to be accessible by the user
4652 under which the QEMU process is running; however, other processes such as
4653 vdsm need to access the sockets as well, which means adjusting permissions
4656 * cpu_map: Add Icelake model definitions
4658 These CPU models will be available in the upcoming 3.1.0 QEMU release.
4660 * util: Properly parse URIs with missing trailing slash
4662 Some storage URIs were not parsed correctly, in which case libvirt ended up
4663 emitting XML that it would then refuse to parse back.
4671 * Xen: Support PM Suspend and Wakeup
4673 The libxl driver now supports the virDomainPMSuspendForDuration and
4674 virDomainPMWakeup APIs.
4676 * **Removed features**
4678 * Xen: Drop support for Xen 4.4 and 4.5
4680 Xen 4.4 and 4.5 are no longer supported by the Xen community. Drop support
4681 for these older versions and require Xen >= 4.6.
4683 * nwfilter: Disallow binding creation in session mode
4685 Ensure that a filter binding creation is not attempted in session mode and
4686 generates a proper error message.
4690 * qemu: Retrieve guest hostname through QEMU Guest Agent command
4692 QEMU is now able to retrieve the guest hostname using a new QEMU-GA command
4693 called 'guest-get-host-name'. Virsh users can execute 'domhostname' for
4694 QEMU driver for domains configured to use the Guest Agent.
4696 * virsh: Implement vsh-table in virsh and virsh-admin
4698 The new API fixes problems with table-alignment, making the tables more
4699 readable and deals with unicode.
4703 * storage: Allow inputvol to be encrypted
4705 When creating a storage volume based on another volume, the base input
4706 volume is allowed to be encrypted.
4708 * virsh: Require explicit --domain for domxml-to-native
4710 The --domain option for domxml-to-native virsh command has always been
4711 documented as required, but commit v4.3.0-127-gd86531daf2 accidentally made
4714 * lxc_monitor: Avoid AB / BA lock race
4716 A deadlock situation could occur when autostarting a LXC domain 'guest' due
4717 to two threads attempting to take opposing locks while holding opposing
4718 locks (AB BA problem).
4726 * storage: add storage pool iscsi-direct
4728 Introduce a new storage pool backend that uses libiscsi instead of
4729 iscsiadm. It support basic pool operations: checkPool and refreshPool.
4731 * Add support for MBA (Memory Bandwidth Allocation technology)
4733 Domain vCPU threads can now have allocated some parts of host memory
4734 bandwidth by using the ``memorytune`` element in ``cputune`` .
4736 * qemu: Add support for RISC-V guests
4738 riscv32 and riscv64 guest architectures are now supported.
4742 * qemu: Add ccw support for vhost-vsock
4744 Support the vhost-vsock-ccw device on S390.
4746 * qemu: Make default machine type independent of QEMU
4748 We can't control whether or not QEMU will change its default machine type
4749 in the future, or whether downstream distributions will decide to compile
4750 out some machine types, so our only option to provide a predictable
4751 behavior is taking care of the default ourselves; management applications
4752 and users are encouraged to explicitly pick a machine type when creating
4755 * apparmor: Various improvements
4757 Rules have been added to deal with a number of scenarios that didn't work
4762 * esx: Truncate CPU model name
4764 Some CPU model names are too long to be stored into the corresponding
4765 property, and should be explicitly truncated to avoid unexpected behavior
4766 in users of the ``virNodeGetInfo()`` API such as ``virsh nodeinfo`` .
4768 * utils: Remove arbitrary limit on socket_id/core_id
4770 Both values were assumed to be smaller than 4096, but in fact they are
4771 entirely hardware-dependent and there have been reports of machines
4772 presenting much bigger values, preventing libvirt from working correctly;
4773 all such limits have now been removed.
4781 * qemu: Implement the HTM pSeries feature
4783 Users can now decide whether HTM (Hardware Transactional Memory) support
4784 should be available to the guest.
4786 * qemu: Enable VNC console for mediated devices
4788 Host devices now support a new attribute 'display' which can be used to
4789 turn on frame buffer rendering on a vgpu mediated device instead of on an
4790 emulated GPU, like QXL.
4794 * qemu: Introduce a new video model of type 'none'
4796 Introduce a new video model type that disables the automatic addition of a
4797 video device to domains with 'graphics' specified in their XML. This can be
4798 useful with GPU mediated devices which can serve as the only rendering
4799 devices within the guest.
4801 * virsh: Add --alias to attach-disk and attach-interface commands
4803 Add option --alias to set customized device alias name when using
4804 attach-disk or attach-interface commands.
4806 * virsh: Support usb and sata address to attach-disk
4808 Usb or sata address could be used when attach-disk with --address. For
4809 example, use usb address as usb:<bus>.<port>, use sata address as
4810 <controller>.<bus>.<unit>.
4818 * qemu: Provide TPM emulator support
4820 Support QEMU's TPM emulator based on swtpm. Each QEMU guest gets its own
4823 * bhyve: Support specifying guest CPU topology
4825 Bhyve's guest CPU topology could be specified using the ``<cpu><topology
4826 ../></cpu>`` element.
4828 * qemu: Add support for extended TSEG size
4830 Support specifying extended TSEG size for SMM in QEMU.
4832 * qemu: Add support for SEV guests
4834 SEV (Secure Encrypted Virtualization) is a feature available on AMD CPUs
4835 that encrypts the guest memory and makes it inaccessible even to the host
4838 * **Removed features**
4840 * Remove support for qcow/default encrypted volumes
4842 Disallow using a qcow encrypted volume for the guest and disallow creation
4843 of the qcow or default encrypted volume from the storage driver. Support
4844 for qcow encrypted volumes has been phasing out since QEMU 2.3 and by QEMU
4845 2.9 creation of a qcow encrypted volume via qemu-img required usage of
4846 secret objects, but that support was never added to libvirt.
4848 * Make GnuTLS mandatory
4850 Building without GnuTLS is no longer possible.
4852 * qemu: Remove allow_disk_format_probing configuration option
4854 The option represented a security risk when used with malicious disk
4855 images, so users were recommended against enabling it; with this release,
4856 it's been removed altogether.
4860 * capabilities: Provide info about host IOMMU support
4862 Capabilities XML now provide information about host IOMMU support.
4864 * virsh: Add --all to domblkinfo command
4866 Alter the ``domblkinfo`` command to add the option --all in order to
4867 display the size details of each domain block device from one command in a
4870 * qemu: Allow concurrent access to monitor and guest agent
4872 Historically libvirt prevented concurrent accesses to the qemu monitor and
4873 the guest agent. Therefore two independent calls (one querying the monitor
4874 and the other querying guest agent) would serialize which hurts
4875 performance. The code was reworked to allow two independent calls run at
4878 * qemu: Allow configuring the page size for HPT pSeries guests
4880 For HPT pSeries guests, the size of the host pages used to back guest
4881 memory and the usable guest page sizes are connected; the new setting can
4882 be used to request that a certain page size is available in the guest.
4884 * Add support to use an raw input volume for encryption
4886 It is now possible to provide a raw input volume as input for to generate a
4887 luks encrypted volume via either virsh vol-create-from or
4888 virStorageVolCreateXMLFrom.
4890 * qemu: Add support for vsock hot (un)plug and cold (un)plug
4892 * qemu: Add support for NBD over TLS
4894 NBD volumes can now be accessed securely.
4896 * qemu: Implement FD passing for Unix sockets
4898 Instead of having QEMU open the socket and then connecting to it, which is
4899 inherently racy, starting with QEMU 2.12 we can open the socket ourselves
4900 and pass it to QEMU, avoiding race conditions.
4902 * virsh: Introduce --nowait option for domstat command
4904 When this option is specified, virsh will try to fetch the guest stats but
4905 abort instead of stalling if they can't be retrieved right away.
4909 * qemu: Fix a potential libvirtd crash on VM reconnect
4911 Initialization of the driver worker pool needs to come before libvirtd
4912 trying to reconnect to all machines, since one of the QEMU processes migh
4913 have already emitted events which need to be handled prior to us getting to
4914 the worker pool initialization.
4916 * qemu: Fix domain resume after failed migration
4918 Recent versions of QEMU activate block devices before the guest CPU has
4919 been started, which makes it impossible to roll back a failed migration.
4920 Use the ``late-block-activate`` migration capability if supported to avoid
4923 * vmx: Permit guests to have an odd number of vCPUs
4925 An odd number of vCPUs greater than 1 was forbidden in the past, but
4926 current versions of ESXi have lifted that restriction.
4934 * bhyve: Support locking guest memory
4936 Bhyve's guest memory may be wired using the
4937 ``<memoryBacking><locked/></memoryBacking>`` element.
4939 * qemu: Provide VFIO channel I/O passthrough support
4941 Support passthrough devices that use channel I/O based mechanism in a QEMU
4944 * qemu: Add support for migration of VMs with non-shared storage over TLS
4946 It's now possible to use the VIR_MIGRATE_TLS flag together with
4947 VIR_MIGRATE_NON_SHARED_DISK. The connection is then secured using the TLS
4948 environment which is setup for the migration connection.
4950 * Add support for VM Generation ID
4952 The VM Generatation ID exposes a 128-bit, cryptographically random, integer
4953 value identifier, referred to as a Globally Unique Identifier (GUID) to the
4954 guest in order to notify the guest operating system when the virtual
4955 machine is executed with a different configuration. Add a new domain XML
4956 processing and a domain capabilities feature.
4958 * Introduce virDomainDetachDeviceAlias
4960 This new API enables users to detach device using only its alias.
4962 * Introduce new virConnectCompareHypervisorCPU and
4963 virConnectBaselineHypervisorCPU APIs
4965 Unlike the old virConnectCompareCPU and virConnectBaselineCPU APIs, both
4966 new APIs consider capabilities of a specific hypervisor.
4968 * Introduce SCSI persistent reservations support
4970 The QEMU driver gained support for qemu-pr-helper which enables guests to
4971 issue SCSI commands for persistent reservation.
4973 * qemu: Implement multiple screen support for virDomainScreenshot
4975 While the virDomainScreenshot API supported multihead video cards, the
4976 implementation was missing. But now that QEMU implemented it libvirt has
4979 * qemu: add support for vhost-vsock-device
4981 A new vsock device was introduced, allowing communication between the guest
4982 and the host via the AF_VSOCK family.
4986 * qemu: Add support for OpenGL rendering with SDL
4988 Domains using SDL as a graphics backend will now be able to use OpenGL
4989 accelerated rendering.
4991 * qemu: Add support for 'output' audio codec
4993 Support QEMU's 'hda-output' codec advertising only a line-out for ich6 and
4996 * virsh: Enhance event name completion
4998 Implement event name completion for some commands (e.g. event,
4999 secret-event, pool-event and nodedev-event)
5007 * qemu: Add support for the pcie-to-pci-bridge controller
5009 Pure PCIe guests such as x86_64/q35 and aarch64/virt will now add this
5010 controller when traditional PCI devices are in use.
5012 * Xen: Support setting CPU features for host-passthrough model
5014 The CPU model presented to Xen HVM domains is equivalent to libvirt's
5015 host-passthrough model, although individual features can be enabled and
5016 disabled via the cpuid setting. The libvirt libxl driver now supports
5017 enabling and disabling individual features of the host-passthrough CPU
5020 * **Removed features**
5022 * Xen: Drop the legacy xend-based driver
5024 The xm/xend toolstack was deprecated in Xen 4.2 and removed from the Xen
5025 sources in the 4.5 development cycle. The libvirt driver based on xend is
5026 now removed from the libvirt sources.
5030 * qemu: Support hot plug and hot unplug of mediated devices
5032 Libvirt now allows mediated devices to be hot plugged and hot unplugged
5033 from a guest rather than reporting an error that this isn't supported. In
5034 fact, kernel has been supporting this since 4.10.
5038 * Improve handling of device mapper targets
5040 When starting a domain with a disk backed by a device mapper volume libvirt
5041 also needs to allow the storage backing the device mapper in CGroups. In
5042 the past kernel did not care, but starting from 4.16 CGroups are consulted
5043 on each access to the device mapper target.
5051 * Support building with Python 3
5053 Python is required to build libvirt, and up until now only Python 2 could
5054 be used as an interpreter. All scripts used during build have now been made
5055 compatible with Python 3, which means both major releases of the language
5056 are fully supported.
5058 * qemu: Provide ccw address support for graphics and input devices
5060 Support the virtio-gpu-ccw device as a video device and virtio-{keyboard,
5061 mouse, tablet}-ccw devices as input devices on S390.
5065 * qemu: Add logging of guest crash information on S390
5067 On S390, when the guest crashes and QEMU exposes the guest crash
5068 information, log the relevant data to the domain log file.
5070 * qemu: use arp table of host to get the IP address of guests
5072 Find IP address of a VM by arp table on hosts. If someone customizing IP
5073 address inside VM, it will be helpful.
5075 * Xen: Remove hard-coded scheduler weight
5077 The libxl driver was accidentally hard-coding the per-domain scheduler
5078 weight to 1000, silently ignoring any user-provided ``<shares>`` in
5079 ``<cputune>`` . The driver now honors ``<shares>`` , and defers setting a
5080 default value to Xen. Note that the Xen default is 256, so any domains
5081 started after this improvement will have one fourth the shares of
5082 previously started domains. If all domains must have equal CPU shares,
5083 administrators must manually set the weight of previously started domains
5084 to 256, or restart them.
5088 * qemu: TLS migration now enforces use of TLS for the NBD connection
5090 When the VIR_MIGRATE_TLS flag was used with the migration API libvirt did
5091 not ensure that the NBD connection was using TLS as well. The code now
5092 rejects such migration as the TLS transport for NBD is not ready yet, but
5093 prevents a false sense of security that TLS would be used. The support TLS
5094 for NBD will be added soon.
5102 * Added support for CAT (Cache allocation Technology)
5104 Domain vCPU threads can now have allocated some parts of host cache using
5105 the ``cachetune`` element in ``cputune`` .
5107 * Allow opening secondary drivers
5109 Up until now it was possible to connect to only hypervisor drivers (e.g.
5110 qemu:///system, lxc:///, vbox:///system, and so on). The internal drivers
5111 (like network driver, node device driver, etc.) were hidden from users and
5112 users could use them only indirectly. Starting with this release new
5113 connection URIs are accepted. For instance network:///system,
5114 storage:///system and so on.
5116 * virtlogd, virtlockd: Add support for admin protocol
5118 These two daemons now support admin protocol through which some admin info
5119 can be gathered or some configuration tweaked on the fly.
5123 * virsh: Enhance bash completion
5125 Implement more bash completions so that basic libvirt objects can be
5126 auto-completed (e.g. networks, interfaces, NWFilters, and so on).
5128 * qemu: Use VIR_ERR_DEVICE_MISSING for various hotplug/detach messages
5130 * qemu: Allow showing the dump progress for memory only dump
5132 Alter the QEMU dump-guest-memory command processing to check for and allow
5133 asynchronous completion which then allows for the virsh dump --memory-only
5134 --verbose command to display percent completion data.
5136 * conf: add support for setting Chassis SMBIOS data fields
5138 * libxl: add support for setting clock offset and adjustment
5140 * Make port allocator global
5142 Up until now each driver had their own port allocator module. This meant
5143 that info on port usage was not shared. Starting with this release, the
5144 port allocator module is made global and therefore drivers allocate ports
5147 * Fixed some compiler warnings that appear with GCC 8
5151 * qemu: Check for unsafe migration more thoroughly
5153 If a domain disk is stored on local filesystem (e.g. ext4) but is not being
5154 migrated it is very likely that domain is not able to run on destination.
5155 Regardless of share/cache mode.
5157 * qemu: Fix updating device with boot order
5159 Starting with 3.7.0 release updating any device with boot order would fail
5160 with 'boot order X is already used by another device' while in fact it was
5161 the very same device.
5163 * virlog: determine the hostname on startup CVE-2018-6764
5165 At later point it might not be possible or even safe to use getaddrinfo().
5166 It can in turn result in a load of NSS module which can even be loaded from
5167 unsage guest filesystem and thus escape the confinment of its container.
5169 * qemu: Rework vCPU statistics fetching
5171 Fetching vCPU statistics was very expensive because it lead to waking up
5172 vCPU threads in QEMU and thus it degraded performance. The code was
5173 reworked so that fetching statistics does not wake up halted vCPUs.
5175 * qemu: unlink memory backing file on domain shutdown
5177 Depending on the filesystem where domain memory is stored, some files might
5178 have been left behind. This is not a problem on hugetlbfs, but it is a
5179 problem on regular filesystems like ext4.
5181 * qemu: Fix shutting down domains in parallel
5183 If multiple domains were being shut down in parallel, libvirtd might have
5186 * nodedev: Update PCI mdev capabilities dynamically
5188 PCI devices may have other nested capabilities, like SRIOV and mdev which
5189 depend on the device being plugged into the native vendor driver. However,
5190 in case such a device is directly assigned to a guest using VFIO driver,
5191 the device will naturally lose these capabilities and libvirt needs to
5200 * tools: Provide bash completion support
5202 Both ``virsh`` and ``virt-admin`` now implement basic bash completion
5205 * qemu: Refresh capabilities on host microcode update
5207 A microcode update can cause the CPUID bits to change; therefore, the
5208 capabilities cache should be rebuilt when such an update is detected on the
5211 * lxc: Set hostname based on container name
5215 * CPU frequency reporting improvements
5217 The CPU frequency will now be reported by ``virsh nodeinfo`` and other
5218 tools for s390 hosts; at the same time; CPU frequency has been disabled on
5219 aarch64 hosts because there's no way to detect it reliably.
5221 * libxl: Mark domain0 as persistent
5223 * Xen: Add support for multiple IP addresses on interface devices
5225 * qemu: Add support for hot unplugging redirdev device
5229 * qemu: Enforce vCPU hotplug granularity constraints
5231 QEMU 2.7 and newer don't allow guests to start unless the initial vCPUs
5232 count is a multiple of the vCPU hotplug granularity, so validate it and
5233 report an error if needed.
5236 v3.10.0 (2017-12-04)
5237 ====================
5241 * conf: Support defining distances between virtual NUMA cells
5243 A NUMA hardware architecture supports the notion of distances between NUMA
5244 cells. This can now be specified using the ``<distances>`` element within
5245 the NUMA cell configuration. Drivers which support this include Xen and
5248 * Xen: Support defining vNUMA topology
5250 Xen now supports defining a virtual NUMA topology for VMs, including
5251 specifying distances between NUMA cells.
5253 * qemu: Add the ability to configure HPT resizing for pSeries guests
5255 The user can now decide whether HPT (Hash Page Table) resizing should be
5256 enabled, disabled or required instead of leaving it up to hypervisor
5257 defaults and negotiation between the guest and the host.
5259 * qemu: Add vmcoreinfo feature
5261 Starting with QEMU 2.11, the guest can save kernel debug details when this
5262 feature is enabled and the kernel supports it. It is useful to process
5263 kernel dump with KASLR enabled, and also provides various kernel details to
5266 * conf: Move the auth and encryption definitions to disk source
5268 Allow parsing and formatting of the ``auth`` and ``encryption``
5269 sub-elements to be a child of the ``source`` element. This will allow
5270 adding an ``auth`` sub-element to a ``backingStore`` or ``mirror`` elements
5271 as a means to track specific authentication and/or encryption needs.
5275 * vbox: Add VirtualBox 5.2 support
5277 * vbox: Add support for configuring storage controllers
5279 The VirtualBox driver now supports the ``<controller>`` element in the
5280 domain XML for configuring storage controllers in VBOX VMs. Additionally,
5281 libvirt's domain XML schema was updated to allow optional ``model``
5282 attribute for ``<controller type='ide'>`` which is used by the VBOX driver
5283 to set the IDE controller model to be one of 'piix4', 'piix4' (default), or
5284 'ich6'. Finally, with this change ``dumpxml`` generates ``<controller>``
5285 elements that correspond to current VBOX VM storage controller
5288 * vbox: Add support for attaching empty removable disks
5290 The VirutalBox driver now supports adding CD-ROM and floppy disk devices
5291 that do not have the disk source specified. Previously such devices were
5294 * vbox: Add support for attaching SAS storage controllers
5296 In VirtualBox, SCSI and SAS are distinct controller types whereas libvirt
5297 does not make such distinction. Therefore, the VBOX driver was updated to
5298 allow attaching SAS controllers via ``<controller type='scsi'
5299 model='lsisas1068'>`` element. If there are both SCSI and SAS controllers
5300 present in the VBOX VM, the domain XML can associate the disk device using
5301 the ``<address>`` element with the ``controller`` attribute, and
5302 optionally, set the port via ``unit`` attribute.
5304 * qemu: Generate predictable paths for qemu memory backends
5306 In some cases management applications need to know paths passed to
5307 memory-backend-file objects upfront. Libvirt now generates predictable
5308 paths so applications can prepare the files if they need to do so.
5310 * Shareable disks work properly with recent qemu
5312 Recent qemu versions added image locking to avoid potential corruption of
5313 disk images. This broke shareable disks with libvirt since the feature was
5314 turned on by default in qemu. Libvirt now enables sharing of those disks in
5315 qemu so that the image locking is not applied in that case. Additionally
5316 libvirt now checks that shareable disks have supported format (raw) to
5317 avoid metadata corruption.
5319 * Improve serial console behavior on non-x86 architectures
5321 ppc64, aarch64 and s390x guests were treating the <serial> and <console>
5322 elements differently from x86, in some cases presenting misleading
5323 information to the user. The behavior is now consistent across all
5324 architectures and the information reported is always accurate.
5328 * vbox: Do not ignore failures to attach disk devices when defining
5330 The ``define`` now fails and reports an error if any of the ``controller``
5331 or ``disk`` devices specified in the domain XML fail to attach to the
5334 * vbox: Fix dumpxml to always output disk devices
5336 The VirtualBox driver was ignoring any disk devices in ``dumpxml`` output
5337 if there was a SAS storage controller attached to the VM.
5339 * vbox: Fix dumpxml to always generate valid domain XML
5341 When a VirtualBox VM has multiple disks attached, each to a different
5342 storage controller that uses 'sd' prefix for block device names e.g. one
5343 disk attached to SATA and one to SCSI controller, it no longer generates
5344 XML where both would have 'sda' device name assigned. Instead it properly
5345 assigns 'sda' and 'sdb' to those disks in the order of appearance.
5347 * Securely pass iSCSI authentication data
5349 Rather than supplying the authentication data as part of the iSCSI URL for
5350 a disk or host device, utilize the encrypted secret object to securely pass
5351 the authentication data.
5359 * Add capability to allow hot (un)plug of a domain watchdog device
5361 * Allow users to set device aliases
5363 Users can set aliases to domain devices and thus identify them easily.
5365 * qemu: Support multiqueue for virtio-blk
5367 Multiqueue support for ``virtio-blk`` has been available in QEMU ever since
5368 2.7.0, and now libvirt guests can enable it.
5370 * Add virDomainSetLifecycleAction API
5372 Provided a new API to allow dynamic guest lifecycle control for guest
5373 reactions to poweroff, restart, or crash type events related to the domain
5374 XML ``on_poweroff`` , ``on_reboot`` , and ``on_crash`` elements. The
5375 ``virsh set-lifecycle-action`` command was created to control the actions.
5377 * qemu: Allow cold(un)plugging and hot(un)plugging input devices
5379 * net: Implement QoS for vhostuser
5383 * Allow a logical volume to be create using LUKS
5385 A logical volume may be created using an ``encryption`` element using
5386 "luks" format. This does require a previously created ``secret`` to store
5387 the passphrase used to encrypt the volume Adding the volume to a domain can
5388 then either provide the secret or allow the consumer in the guest to
5389 provide the passphrase in order to decrypt the volume.
5391 * net: Ignore auto-generated MAC address when detaching an interface
5393 If the MAC address has not been specified by the user, libvirt will try and
5394 fill in the gaps by generating one; however, for some error paths that led
5395 to some confusing error messages, so when an auto-generated MAC address is
5396 specified the error message will not include the auto-generated MAC.
5398 * net: Enable MAC address lookup for virDomainInterfaceStats
5400 * apparmor: Several improvements
5402 Changes include permitting access to data about USB devices and ``dnsmasq``
5403 instances, allowing spaces in guest names and many more.
5405 * cpu: Use CPU information obtained from QEMU when possible
5407 Recent QEMU versions can expose information about which CPU models are
5408 available and usable on the host; libvirt will now make use of such
5409 information whenever possible.
5411 * hyperv: Various improvements
5413 The error reported when clients can't connect to Hyper-V has been made more
5414 descriptive, and memory limits for guests are now mapped to more
5415 appropriate libvirt equivalents.
5417 * qemu: Report QEMU error on failed migration
5419 Instead of reporting a generic error, ask QEMU for a more detailed and thus
5420 hopefully more helpful one.
5422 * vbox: Implement autoport for RDP
5424 libvirt will now obtain the (dynamically allocated) RDP port number from
5425 VirtualBox itself, avoiding conflicts between multiple guests wanting to
5426 use RDP at the same time.
5428 * qemu: Allow rotation of small logs
5430 On a host where numerous unique instances are executed per day, it's quite
5431 possible that, even though each of the single log files are fairly small,
5432 collectively the quantity and volume may add tens of thousands of log files
5433 to the ``/var/log/libvirt/qemu/`` directory. Removing the constraints that
5434 log have to be bigger than 100 KiB before they can be rotated solves the
5439 * Fix swapped interface statistics and QoS
5441 Due to internal implementation, reported statistics for some types of
5442 interfaces were swapped (RX appeared in TX and vice versa). Similarly, QoS
5443 was set in reversed way.
5445 * Properly resize local LUKS encrypted volume
5447 Resizing of a local LUKS encrypted volume will now use qemu-img to resize
5448 the volume. This will require configuring a secret for the LUKS encrypted
5451 * qemu: Reserve PCI addresses for implicit i440fx devices
5453 Failing to do so causes the addresses to be considered usable by libvirt,
5454 which means they could be assigned to more than one device resulting in the
5455 guest failing to start.
5457 * spec: Restart libvirtd only at the end of the upgrade process
5459 Use ``%posttrans`` to make sure ``libvirtd`` is not restarted before all
5460 other components, such as the library itself and storage / hypervisor
5461 drivers, have already been upgraded.
5465 * qemu: Ensure TLS clients always verify the server certificate
5467 While it's reasonable to turn off client certificate validation, as setting
5468 it up can be non-trivial, clients should always verify the server
5469 certificate to avoid MITM attacks. However, libvirt was using the same knob
5470 to control both checks, leading to CVE-2017-1000256 / LSN-2017-0002.
5478 * qemu: Added support for cold-(un)plug of watchdog devices
5480 * qemu: Added support for setting IP address os usernet interfaces
5482 * qemu: Added support for Veritas Hyperscale (VxHS) block devices
5484 * storage: Added new events for pool-build and pool-delete
5488 * qemu: Set DAC permissions properly for spice rendernode
5490 When a ``rendernode`` path is set for SPICE GL on ``qemu:///system`` , we
5491 now correctly set DAC permissions on the device at VM startup. This is the
5492 last remaining hurdle to let SPICE GL work for ``qemu:///system`` without
5493 any external host changes.
5495 * nodedev: Add switchdev offload query to NIC capabilities
5497 Allow querying the NIC interface capabilities for the availability of
5498 switchdev offloading (also known as kernel-forward-plane-offload).
5500 * New CPU models for AMD and Intel
5502 AMD EPYC and Intel Skylake-Server CPU models were added together with their
5505 * Improve long waiting when saving a domain
5507 While waiting for a write to disk to be finished, e.g. during save, even
5508 simple operations like ``virsh list`` would be blocking due to domain lock.
5509 This is now resolved by unlocking the domain in places where it is not
5514 * Proper units are now used in virsh manpage for dom(mem)stats
5516 Previously the documentation used multiples of 1000, but now it is fixed to
5517 use multiples of 1024.
5519 * qemu: Fix error reporting when disk attachment fails
5521 There was a possibility for the actual error to be overridden or cleared
5522 during the rollback.
5524 * qemu: Fix assignment of graphics ports after daemon restart
5526 This could be seen with newer kernels that have bug regarding SO_REUSEADDR.
5527 After libvirtd was restarted it could assign already used address to new
5528 guests which would make them fail to start. This is fixed by marking used
5529 ports unavailable when reconnecting to running QEMU domains.
5531 * Fix message decoding which was causing a very strange bug
5533 When parsing an RPC message with file descriptors was interrupted and had
5534 to restart, the offset of the payload was calculated badly causing strange
5535 issues like not being able to find a domain that was not requested.
5543 * qemu: Add managedsave-edit commands
5545 Using managedsave-dumpxml, managedsave-define and managedsave-edit
5546 commands, now we can dump and edit the XML configuration of domain which
5547 has managedsave image.
5549 * qemu: Add migrate-getmaxdowntime command
5551 Currently, the maximum tolerable downtime for a domain being migrated is
5552 write-only from libvirt, via migrate-setmaxdowntime. This implements a
5553 complementary migrate-getmaxdowntime command
5555 * bhyve: Support autoport for VNC ports
5557 It's no longer necessary to explicitly specify VNC port for the bhyve
5558 guests. With the autoport feature it will be allocated automatically.
5559 Please refer to the bhyve driver documentation for examples.
5561 * qemu: Added support for setting heads of virtio GPU
5563 * qemu: Added support to configure reconnect timeout for chardev devices
5565 When you have a TCP or UNIX chardev device and it's connected somewhere you
5566 can configure reconnect timeout if the connection is closed.
5570 * qemu: Report a clear error when dropping a VM during startup
5572 "Failed to load config for domain 'DOMNAME'" is now reported if a VM config
5573 can't be parsed for some reason, and thus provides a clear indication for
5576 * apparmor: Update for QEMU 2.10 compatibility
5578 Starting with QEMU 2.10, disk images and NVRAM files get automatically
5579 locked to prevent them from being corrupted; however, file locking needs to
5580 be explicitly allowed through ``virt-aa-helper`` or AppArmor will reject
5581 the requests and the guest will not be able to run.
5583 * virsh: List Unix sockets in 'domdisplay' output
5585 VNC and SPICE graphics can use Unix sockets instead of TCP/IP sockets as
5586 connection endpoints, but such a configuration was not handled correctly by
5587 ``virsh domdisplay`` , causing the respective endpoints to be missing from
5590 * qemu: Don't check whether offline migration is safe
5592 Since offline migration only copies the guest definition to the destination
5593 host, data corruption is not a concern and the operation can always be
5596 * virt-host-validate: Fix IOMMU detection on ppc64
5600 * qemu: Better support for international domain names (with wide characters)
5602 There were some issues with multi-byte domains getting lost on daemon
5603 restart due to truncation, so the code now handles multi-byte names a bit
5606 * qemu: Support long domain names with namespaces
5608 Domains with extremely long names would fail to start due to temporary
5609 namespace paths being created with the whole name. The path is now
5610 generated with shortened name instead.
5612 * qemu: Tolerate missing emulator binary during libvirtd restart
5614 For some time libvirt required qemu capabilities being present when parsing
5615 VM configs during startup. As a side effect VM configs would fail to parse
5616 and thus vanish, if the emulator binary would be uninstalled or broken.
5617 Libvirt now tolerates when capabilities are missing during startup.
5619 * qemu: Prevent pSeries guests from disappearing in some situations
5621 pSeries guest would disappear if any of the host devices they were
5622 configured to use was not available during libvirtd startup, which could
5623 easily happen for SR-IOV Virtual Functions. This scenario is now handled
5626 * qemu: Honor <on_reboot/> setting
5628 The setting was accepted by the parser, but not actually implemented.
5630 * Fix --verbose option for all daemons
5632 Since v3.0.0, the option had been ignored by all libvirt daemons (
5633 ``libvirtd`` , ``virtlogd`` and ``virtlockd`` ); it's now working as
5634 intended once again.
5642 * hyperv: Implement virDomainSetMemory and virDomainSendKey APIs
5644 * qemu: Support multiple PHBs for pSeries guests
5646 pSeries guests can now have multiple PHBs (PCI Host Bridges), which show up
5647 as separate PCI domains in the guest. To create additional PHBs, simply add
5648 PCI controllers with model ``pci-root`` to the guest configuration.
5650 * qemu: Isolate hostdevs on pSeries guests
5652 To enable better error reporting and recovery, unrelated hostdevs will now
5653 be automatically isolated on pSeries guests by placing them on separate
5654 PHBs (PCI Host Bridges).
5658 * qemu: platform serial devices can now use chardev features
5660 QEMU VMs that depend on platform serial devices can now use QEMU's
5661 ``-chardev`` option, which enables access to advanced features like log
5662 file configuration. This applies to the default serial devices for arm,
5663 aarch64, and some ppc configurations.
5665 * Require use of GCC 4.4 or Clang compilers
5667 We only ever test libvirt with GCC or Clang (which provides a GCC
5668 compatible compilation environment). Between them, these compilers cover
5669 every supported operating system platform, including Windows.
5671 * qemu: shared disks with directsync cache should be safe for migration
5673 At present shared disks can be migrated with either readonly or
5674 ``cache=none`` . But ``cache=directsync`` should be safe for migration,
5675 because both ``cache=directsync`` and ``cache=none`` don't use the host
5676 page cache, and ``cache=direct`` write through qemu block layer cache.
5678 * Handle hotplug change on VLAN configuration using OVS
5680 Libvirt was taught to handle VLAN change for running OVS interface.
5684 * qemu: Use vCPU 'node-id' property and pass it back to qemu
5686 vCPU properties gathered from query-hotpluggable-cpus need to be passed
5687 back to QEMU. As QEMU did not use the node-id property until now and
5688 libvirt forgot to pass it back properly (it was parsed but not passed
5689 around) we did not honor this.
5691 * Miscellaneous stream fixes
5693 After introducing sparse stream features there were still some known bugs
5694 left. Those are fixed in this release.
5696 * qemu: Miscellaneous domain NS fixes
5698 Libvirt starts qemu domains in separate Linux namespaces for a while now.
5699 However, there were still some bugs lingering. For instance libvirt did not
5700 know how to handle file based bind mounts.
5702 * Various CPU driver improvements
5704 There were some minor bugs when using 'host-model' CPU.
5712 * qemu: Add support for loadparm for a boot device
5714 Add an optional boot parameter 'loadparm' for a boot device. Loadparm is an
5715 8 byte parameter that, when present, is queried by S390 guests via sclp or
5716 diag 308. Linux guests on S390 use it to select a boot entry.
5718 * Support controlling how video devices are exposed to the bhyve guests
5720 The ``vgaconf`` attribute was added to ``video`` 's ``driver`` element.
5721 Possible values are: ``on`` , ``off`` , and ``io`` . It controls the way
5722 how bhyve exposes video devices to its guests; various guest OSes might
5723 require different settings to boot properly.
5725 * qemu: Add support for live updates of coalesce settings
5727 Users can now use ``virsh update-device`` to change the coalesce settings
5728 of an interfaces while the domain is running.
5730 * qemu: Allow VirtIO devices to use vIOMMU
5732 It is now possible to turn on IOTBL for the vIOMMU and have VirtIO devices
5733 use it, provided they have been configured appropriately.
5737 * qemu: block copy job can be used with persistent domains
5739 Until now it was not possible to use block copy with persistent VMs. In use
5740 cases where it's not required to recover the job after VM shutdown, it's
5741 possible to specify VIR_DOMAIN_BLOCK_COPY_TRANSIENT_JOB flag to start the
5744 * JSON pseudo-protocol backing store parser supports new format of qemu 2.9
5746 QEMU 2.9 modified a few structures corresponding to the JSON format of
5747 specifying a backing store for a disk image. Libvirt now implements the new
5750 * Capabilities now include info about host's CAT settings
5752 Various information about resource control from the host is gathered and
5753 presented in capabilities if available.
5755 * apparmor: Several improvements
5757 Allow access to Ceph config, EFI firmware on both x86_64 and aarch64,
5758 device tree on ppc64 and more.
5760 * qemu: Support host-model on POWER9 machines
5764 * qemu: snapshot: retrieve image metadata from user provided files
5766 Disk images of an external snapshot created with
5767 VIR_DOMAIN_SNAPSHOT_CREATE_REUSE_EXT flag specified would not be scanned
5768 for metadata after recent changes. The metadata is necessary to allow
5769 keeping relative paths between images when doing a block-commit.
5771 * Parse decimal numbers in a locale-independent way
5773 Some locales, such as ``de_DE`` and ``pt_BR`` , use comma rather than dot
5774 to separate the integer part from the fractional part of a decimal number;
5775 however, several data sources such as the kernel use a locale-independent
5776 representation and need to be treated accordingly.
5778 * Support compilation with newer compiler and libc versions
5780 Several fixes have been included to make compilation with Clang 4.0.0, GCC
5781 7.1 and glibc >= 2.25.90 possible.
5783 * qemu: Query name for vhost-user interfaces at runtime
5785 This makes it possible to use ``virsh`` subcommands such as ``domiflist``
5786 and ``domifstat`` on vhost-user interfaces.
5788 * qemu: Set MTU for hotplugged interfaces correctly
5790 When hotplugging a network interface, the MTU was only set on the guest
5791 side. Set it on the host side as well.
5793 * qemu: Forbid updating MTU for interfaces of running guests
5795 The MTU setting can't be modified while the guest is running, so any
5796 attempt to alter it at runtime will now result in an error rather than
5797 being silently ignored.
5799 * qemu: Fix specifying QXL heads with older QEMU releases
5801 Specifying the number of QXL heads was not working correctly for QEMU
5802 releases older than 1.6.
5804 * qemu: Fix migration to older libvirt/QEMU versions
5806 When the guest is started, libvirt updates the CPU definition to reflect
5807 the actual CPU features to enforce ABI. We need to send original and
5808 updated CPU definition in order to support migration to older libvirt/QEMU
5809 versions. Only the updated CPU definition was sent to destination.
5817 * Improved streams to efficiently transfer sparseness
5819 New extension to virStream was implemented so that virStorageVolDownload
5820 and virStorageVolUpload can preserve file sparseness.
5822 * I/O APIC type can be specified for QEMU/KVM domains
5824 The ``ioapic`` tag was added to domain ``features`` , so the type of the
5825 I/O APIC can now be specified (e.g. putting it in userspace for KVM
5828 * The reason for VM shutdown is reported, if known
5830 QEMU 2.10 will be able to report the reason for shutting down (whether that
5831 was caused by the guest or not), and libvirt is prepared for that and
5832 reports that information in its shutdown event as well, if it is known.
5836 * Repository now has new README.md file
5838 The new file uses markdown syntax, so it looks better on github and
5839 possibly other web pages, but it has also more useful information. The old
5840 README is now symlink to the new file.
5842 * qemu: Use GICv2 by default for aarch64/virt TCG guests
5844 The emulated GICv3 has some limitations that make it unusable as a default;
5845 use GICv2 until they're sorted out. This change makes it once again
5846 possible to run aarch64/virt guests on a x86_64 host without having to
5847 tweak their configuration.
5849 * Additional capabilities for the node_device module
5851 Introduce two new capabilities to the node_device module. The first is for
5852 CCW devices, most common on the S390 architecture. The second is for fibre
5853 channel-backed SCSI devices and exposes the fc_remote_port sub-capability
5854 to SCSI target devices.
5856 * Node devices now report Mediated device capabilities
5858 Endpoint devices support new ``mdev`` capability and their parents now
5859 report the supported types in new ``mdev_types`` capability.
5861 * Capabilities now report information about host caches
5863 If supported in the kernel, host capabilities will now list L3 caches. The
5864 code for other levels was added as well, but only L3 caches are reported
5867 * POWER9 CPU model was added
5869 It is now properly reported in host capabilities.
5871 * libxl: NUMA sibling distances are now reported in host capabilities
5873 * VMDK version 3 files are now properly detected
5875 * Interrupt remapping and Extended interrupt mode for IOMMU devices
5877 These two new features can now be controlled with new ``<driver
5878 intremap='on/off' eim='on/off'/>`` tag for ``iommu`` devices.
5880 * Graphics in libxl domains now have default addresses
5882 Even though there were default addresses before this change, they were not
5883 saved in the XML. It is now possible to see and control the listen
5886 * Default USB controllers are now added for devices in libxl domains
5888 Even though they were added automatically when USB device was attached,
5889 they could've been missing in some other cases. The logic is now fixed so
5890 there are always USB controllers, even if there was none of them in the
5893 * Limits for RPC messages were increased
5895 Hitting the RPC limits we have is easier every day, so they were increased
5896 once again and some guessing logic was improved as well. It is now possible
5897 to get more stats than ever using the ``virConnectGetAllDomainStats()``
5898 call and push through even bigger requests and replies for all APIs.
5902 * qemu: Create memory_backing_dir on startup
5904 Libvirt's policy is that directories are created on startup if they don't
5905 exist. We've missed this one.
5907 * PCIe 4.0 cards now report proper link speeds
5909 It could happen that the link speed for PCIe devices was not properly
5910 reported or the nodedev-dumpxml just failed. That was due to mistake in the
5911 field width, but should now work properly.
5913 * qemu: Do not report errors on shutdown
5915 For some users, in some rare cases, it could happen that there was an error
5916 message "internal error: End of file from qemu monitor" in the logs even
5917 though no problem happened. The detection of these false positives was
5918 improved and such errors should not show any more.
5920 * User-specified UNIX socket paths for virtio channels should not be reset
5922 It could happen, in some cases, that libvirt would mistake a user-specified
5923 path for its own generated one and thus remove it from the XML. The
5924 detection of such addresses was improved now.
5926 * Fix address reservation during RNG hot-plug
5928 When error occurred in a specific point in time during the hot-plug of an
5929 RNG device, it could happen that an address was released even though
5930 another device was already using it, making it possible to hot-plug another
5931 device with that address, effectively having duplicated addresses in the
5940 * net: Add support for coalesce settings
5942 Enabling data batching through these settings can improve network
5943 performance for guests.
5945 * qemu: Add support for guest CPU cache specification
5947 This features allows fine-grained control of the cache behavior of the
5950 * qemu: Add support for the qemu-xhci USB controller
5954 * hyperv: Support Hyper-V 2012 and newer
5956 Starting with Hyper-V 2012 the API has changed causing the existing driver
5957 to be unable to send and process requests properly. This has been resolved
5958 by adding abstractions to handle the differences and ease handling such
5959 breaks if they happen in the future.
5961 * libxl: Add support for nested HVM domains
5963 Xen has supported nested HVM domains since version 4.4. The libvirt libxl
5964 driver now supports nested HVM domains by specifying the host-passthrough
5965 CPU mode when defining a domain.
5967 * qemu: Implement ACPI support for aarch64 guests
5969 Up until this point, ACPI support was only advertised for x86_64 guests and
5970 disabling it for aarch64 guests was not possible at all.
5972 * vz: Add support for changing the number of vCPUs
5974 * qemu: Automatically choose the best USB controller for guests
5976 The recently introduced qemu-xhci USB controller is the best choice for
5977 both ppc64 and aarch64 guests, so use it by default on those architectures
5980 * daemon: Increase default task limit for libvirtd
5982 The default number of tasks for the pids cgroup controller is 512, which
5983 libvirtd can quickly bump into when starting lots of guests. Raise the
5984 limit to a more reasonable 32768.
5986 * docs: Include man pages describing key code names and values
5988 * virsh: Report initialization errors
5990 Sometimes virsh might be unable to start: when that happens, report useful
5991 diagnostics instead of failing silently.
5995 * nss: Don't require a network restart for libvirt_guest
5997 Previously, the libvirt_guest NSS module would only work properly after the
5998 corresponding network had been restarted; now newly started guests will be
5999 reported correctly right away.
6001 * storage: Remove unavailable transient pools after restart
6003 Solve an issue where transient storage pools would be stuck in an
6004 unmanageable state if the source disappeared and libvirtd was subsequently
6007 * storage: Fix capacity value for LUKS encrypted volumes
6009 The 'capacity' value (e.g. guest logical size) for a LUKS volume is smaller
6010 than the 'physical' value of the file in the file system, so we need to
6013 * qemu: Fix regression when hyperv/vendor_id feature is used
6015 Guests using the feature would not be started at all; it is now possible to
6016 start them as expected.
6018 * qemu: Do not crash on USB address with no port and invalid bus
6020 * crypto: Always pad data before encrypting it
6022 If this step is not performed, when the data length matches the chunk size
6023 the decryption routines will misinterpret the last byte of data as the
6024 padding length and fail to decode it correctly.
6032 * The virt-host-validate tool now supports bhyve hypervisor
6034 * Introduce NVDIMM memory model
6036 NVDIMM is new type of memory introduced into QEMU 2.6. The idea is that we
6037 have a non-volatile memory module that keeps the data persistent across
6038 domain reboots and offers much faster data accesses. However, due to a bug
6039 in QEMU, this feature is not enabled for QEMUs older than 2.9.0.
6041 * qemu: Introduce support for generic PCIe Root Ports
6043 For new controllers, a generic device (pcie-root-port) will be used by
6044 default instead of the Intel-specific device (ioh3420), provided the QEMU
6047 * qemu: Add support for checking guest CPU ABI compatibility
6049 When migrating a domain to a different host, restoring a domain from a file
6050 or reverting a snapshot libvirt will make sure the guest CPU QEMU presents
6051 to the guest OS exactly matches the one provided on the source host (or
6052 before the domain's state was saved). This enhanced check may also be
6053 requested when starting a new domain to ensure the virtual CPU exactly
6054 matches the one specified in the XML.
6056 * qemu: Add support to migrate using TLS
6058 Add the ability to migrate QEMU guests using TLS via a new flag
6059 VIR_MIGRATE_TLS or virsh migrate '--tls' option. Requires using at least
6060 QEMU 2.9.0 in order to work properly.
6062 * qemu: add mediated devices framework support
6064 Recent kernel version introduced new mediated device framework, so provide
6065 an initial support of this framework for libvirt, mainly by introducing a
6066 new host device type in the XML.
6068 * qemu: Add support for setting TSC frequency
6070 Setting TSC frequency is required to enable migration for domains with
6071 'invtsc' CPU feature turned on.
6073 * Add support for block device threshold event
6075 When using thin provisioning, management tools need to resize the disk in
6076 certain cases. To avoid having them to poll disk usage this version
6077 introduces an event which will be fired when a given offset of the storage
6078 is written by the hypervisor. Together with the API it allows registering
6079 thresholds for given storage backing volumes and this event will then
6080 notify management if the threshold is exceeded. Currently only the qemu
6081 driver supports this.
6083 * bhyve: Add support for UEFI boot ROM, VNC, and USB tablet
6085 The bhyve driver now supports booting using the UEFI boot ROM, so
6086 non-FreeBSD guests that support UEFI could be booted without using an
6087 external boot loader like grub-bhyve. Video is also supported now, allowing
6088 to connect to guests via VNC and use an USB tablet as an input device.
6089 Please refer to the driver page for domain XML examples.
6093 * qemu: Detect host CPU model by asking QEMU on x86_64
6095 Previously, libvirt detected the host CPU model using CPUID instruction,
6096 which caused libvirt to detect a lot of CPU features that are not supported
6097 by QEMU/KVM. Asking QEMU makes sure we don't start it with unsupported
6100 * perf: Add more perf statistics
6102 Add support to get the count of cpu clock time, task clock time, page
6103 faults, context switches, cpu migrations, minor page faults, major page
6104 faults, alignment faults, emulation faults by applications running on the
6107 * Write hyperv crash information into vm log
6109 qemu's implementation of the hyperv panic notifier now reports information
6110 about the crash from the guest os. Starting with this version, libvirt logs
6111 the information to the vm log file for possible debugging.
6115 * QEMU: Use adaptive timeout for connecting to monitor
6117 When starting qemu, libvirt waits for qemu to create the monitor socket
6118 which libvirt connects to. Historically, there was sharp 30 second timeout
6119 after which the qemu process was killed. This approach is suboptimal as in
6120 some scenarios with huge amounts of guest RAM it can take a minute or more
6121 for kernel to allocate and zero out pages for qemu. The timeout is now
6122 flexible and computed by libvirt at domain startup.
6124 * Overwrite (clear) 2 KB instead of just 512 bytes when initializing logical
6127 * Describe the logical backend requirements better for pool-create-as
6135 * storage: Add Virtuozzo storage backend storage pool
6137 Add new storage backend to support pool and volume management within the
6138 Virtuozzo Storage environment. Virtuozzo Storage is a highly available
6139 distributed software defined storage with built-in replication and disaster
6142 * qemu: Add support for memory backing with file source
6144 Add support in numa topology for file source inside memory backing
6145 (hugepages are not needed) Three new elements <source/>,<access/> and
6146 <allocation/> were added to <memoryBacking/> element. Also new
6147 configuration parameter ``memory_backing_dir`` was added to qemu.conf.
6149 * network: make openvswitch call timeout configurable
6151 Adding the ability to specify the timeout value in seconds for openvswitch
6152 calls in the libvirtd configuration file.
6154 * bhyve: add e1000 NIC support
6156 Add support for e1000 NIC. Previously, the only available option was
6159 * libxl: add tunneled migration support
6161 Add tunneled migration to libxl driver, which is always capable of strong
6162 encryption and doesn't require any extra network connection other than
6163 what's required for remote access of libvirtd.
6165 * qemu: add rendernode argument
6167 Add a new attribute 'rendernode' to <gl> spice element.
6169 * nodedev: add drm capability
6171 Add a new 'drm' capability for Direct Rendering Manager (DRM) devices,
6172 providing device type information.
6174 * Add API for individual/specific vCPU hotplug
6176 The new API allows selecting specific vCPUs to be added/removed from the
6177 VM. The existing APIs allowed only adding/removing from the end which did
6178 not play well with NUMA.
6182 * virsh: pool-list: allow both --uuid and --name in one command
6184 Adjust the virsh-pool command to support the --uuid and/or --name options
6185 in order to print just the --name and/or --uuid of pools.
6187 * Introduce MTU to domain <interface/> and <network>
6189 Allow setting MTU size for some types of domain interface and network.
6191 * libxl: improve support for <timer> configurations
6193 Add support for multiple timers. Extend the tsc timer to support the
6194 emulate mode. Improve conversion of timer XML to/from xl.cfg.
6196 * storage: modularize the storage driver
6198 Split up the storage driver backends into loadable modules so that binary
6199 distributions don't have to compromise on shipping the storage driver with
6200 all backends which may pull in too many dependencies.
6204 * nodedev: Fabric name must not be required for fc_host capability
6206 fabric_name is one of many fc_host attributes in Linux that is optional and
6207 left to the low-level driver to decide if it is implemented. For example
6208 the zfcp device driver does not provide a fabric name for an fcp host. The
6209 requirement for the existence of a fabric name has been removed by making
6212 * bhyve: change address allocation schema for SATA disks
6214 Previously, the bhyve driver assigned PCI addresses to SATA disks directly
6215 rather than assigning that to a controller and using SATA addresses for
6216 disks. It was implemented this way because bhyve has no notion of an
6217 explicit SATA controller. However, as this doesn't match libvirt's
6218 understanding of disk addresses, the bhyve driver was changed to follow the
6219 common schema and have PCI addresses for SATA controllers and SATA
6220 addresses for disks. If you're having issues because of this, it's
6221 recommended to edit the domain's XML and remove <address type='pci'> from
6222 the <disk> elements with <target bus='sata'/> and let libvirt regenerate it
6225 * libxl: maximum memory fixes
6227 Fix reporting of domain maximum memory. Fix setting dom0 maximum memory.
6229 * libxl: fix disk detach when <driver> not specified
6231 * libxl: fix dom0 autoballooning with Xen 4.8
6233 * qemu: Allow empty script path to <interface/>
6235 Historically, this was always allowed. Unfortunately, due to some rework
6236 done for 1.3.2 release a bug was dragged in which suddenly stop allowing
6237 domain with such configuration to start.
6245 * Domain events for metadata content changes
6247 The domain events framework has a new event ID that can be used to get
6248 notifications when domain metadata content changes.
6250 * Event notifications for the secret object
6252 The secret object now supports event notifications, covering lifcycle
6253 changes and secret value changes.
6255 * New localPtr attribute for "ip" element in network XML
6257 * qemu: Support QEMU group I/O throttling
6259 Add the capability to allow group I/O throttling via a new domain <disk>
6260 <iotune> subelement "group_name" to allow sharing I/O throttling quota
6261 between multiple drives.
6263 * nss: Introduce libvirt_guest
6265 New ``libvirt_guest`` nss module that translates libvirt guest names into
6268 * daemon: Add support for runtime logging settings adjustment
6270 Logging-related settings like log outputs and filters can now be adjusted
6271 during runtime using the admin interface without the necessity of the
6274 * storage: Add virStorageVolInfoFlags API
6276 Add the API to support using the VIR_STORAGE_VOL_GET_PHYSICAL flag in order
6277 to return the host physical size in bytes of the image container in the
6278 allocation field of the _virStorageVolInfo structure. The --physical flag
6279 has been added to the virsh vol-info command to access the data.
6281 * libxl: Implement virDomainGetMaxVcpus API
6283 * storage: Add overwrite flag checking for logical pool
6285 Add support for the OVERWRITE flags for the logical storage backend
6286 including checking for existing data on the target volumes when building a
6287 new logical pool on target volume(s).
6289 * qemu: Add support for guest CPU configuration on s390(x)
6293 * perf: Add more perf statistics
6295 Add support to get the count of branch instructions executed, branch
6296 misses, bus cycles, stalled frontend cpu cycles, stalled backend cpu
6297 cycles, and ref cpu cycles by applications running on the platform.
6299 * conf: Display <physical> for volume xml
6301 Add a display of the <physical> size of a disk volume in the output of the
6304 * qemu: Use virtio-pci by default for aarch64 mach-virt guests
6306 virtio-pci provides several advantages over virtio-mmio, such as the
6307 ability to hotplug devices and improved performance. While opting in to
6308 virtio-pci has been possible for a while, newly-defined guests will now use
6311 * vbox: remove support for VirtualBox 3.x and older
6313 Those old VirtualBox versions have been unsupported by upstream for a long
6314 time and the API of 4.0 and newer has diverged enough to require code
6315 abstractions to handle differences. Removing support for those old versions
6316 drops lots of code from the driver and simplifies the logic to ease
6317 implementation of new features going forward.
6319 * virsh: pool-info: introduce option --bytes
6321 Add option --bytes to virsh pool-info in order ti allow display of units in
6322 bytes rather than default of human readable output.
6324 * scsi: Add parent wwnn/wwpn or fabric capability for createVport
6326 Improve the algorithm searching for the parent scsi_host device for
6327 vHBA/NPIV scsi_host creation. Rather than supplying the "parent" by name,
6328 it's now possible to define the parent by it's wwnn/wwpn or fabric_wwn in
6329 the node device create XML or the storage pool XML.
6331 * qemu: aggregate pcie-root-ports onto multiple functions of a slot
6333 When pcie-root-ports are added to pcie-root in order to provide a place to
6334 connect PCI Express endpoint devices, libvirt now aggregates multiple root
6335 ports together onto the same slot (up to 8 per slot) in order to conserve
6336 slots. Using this method, it's possible to connect more than 200 endpoint
6337 devices to a guest that uses PCIe without requiring setup of any PCIe
6342 * lxc: fix accidental killing of containers during libvirtd restart
6344 The libvirt_lxc process was previously not moved into the container scope.
6345 As a result, if systemd reloads its config after a container is started,
6346 when libvirtd is later restarted it will accidentally kill the containers.
6348 * qemu: Correct GetBlockInfo values
6350 For an active domain, correct the physical value provided for a raw sparse
6351 file backed storage and the allocation value provided for a qcow2 file
6352 backed storage that hasn't yet been opened on the domain.
6354 * qemu: Make virtio console usable on ppc64 guests
6356 The chardev detection code has been improved and can now handle this
6357 configuration properly.
6359 * qemu: Enable mount namespace
6361 To avoid funny races with udev relabelling devices under our hands and to
6362 enhance security, libvirt now spawns each qemu process with its own
6365 * storage: Fix implementation of no-overwrite for file system backend
6367 Fix file system storage backend implementation of the OVERWRITE flags to be
6368 consistent between code and documentation. Add checks to ensure that when
6369 building a new file system on a target volume that there is not something
6370 already on the disk in a format that libvirt can recognize.
6372 * qemu: Create hugepage path on per domain basis
6374 Historically, all hugepage enabled domains shared the same path under
6375 hugetlbfs. This left libvirt unable to correctly set security labels on it.
6376 With this release, however, each domain is put into a separate path which
6377 is also correctly labeled.
6379 * conf: Reject domains with duplicate drive addresses
6381 Reject duplicate drive addresses for disks and hostdevs at domain
6384 * libxl: reverse defaults on HVM net device attach
6386 Fixes network interface attach for HVM domains when no model is specified.
6387 Emulated hotplug isn't yet supported and hence we should default to the
6388 general working scenario.
6390 * libxl: always enable pae for x86_64 HVM
6392 By default pae is disabled in libxl. Without an explicit <pae/> setting in
6393 the domain <features> configuration, an x86_64 HVM domain would be get an
6394 i686 environment. pae should always be enabled for x86_64 HVM domains.
6396 * qemu: Fix XML dump of autogenerated websocket
6398 As a result autogenerated websocket port is regenerated on domain restore,
6399 migration and such as it should be.
6407 * shmem: Add support for additional models
6409 The shmem device can now utilize QEMU's ivshmem-plain and ivshmem-doorbell,
6410 more modern versions of ivshmem.
6412 * vbox: Add VirtualBox 5.1 support
6414 * libssh: New transport
6416 The new libssh transport allows one to connect to a running libvirtd via
6417 SSH, using the libssh library; for example: ``qemu+libssh://server/system``
6420 * vhost-scsi: Add support scsi_host hostdev passthrough
6422 Add the capability to pass through a scsi_host HBA and the associated LUNs
6425 * Allow debugging of gluster volumes in qemu
6427 Users can now enable debug logging for native gluster volumes in qemu using
6428 the "gluster_debug_level" option in qemu.conf
6430 * Pre-allocate memory slots for memory hotplug
6432 Slot numbers for memory devices are now automatically allocated and thus
6433 persistent. In addition slot numbers can be specified without providing a
6434 base address, which simplifies user configuration
6436 * qemu: Express devices will be placed on PCIe bus by default
6438 For machine types that use a PCI Express root bus (e.g. x86_64/Q35 and
6439 aarch64/virt), any unaddressed PCI device that is an Express device (all
6440 virtio-1.0 devices, e1000e, nec-xhci, vfio assigned devices) will be placed
6441 on an Express controller (i.e. a pcie-root-port) instead of a legacy PCI
6442 controller (i.e. pci-bridge) with the root ports added as needed.
6446 * docs: Better documentation for migration APIs and flags
6448 * vbox: Address thread safety issues
6450 * virsh: Add support for passing an alternative persistent XML to migrate
6453 * vhostuser: Allow hotplug of multiqueue devices
6455 * NEWS: Switch to an improved format
6457 List user-visible changes instead of single commits for a better high-level
6458 overview of differences between libvirt releases.
6460 * website: Modernize layout and branding
6462 The libvirt website looked very cluttered and outdated; it has now been
6463 completely overhauled, resulting in a design that's better organized and
6464 more pleasant to look at.
6468 * vz: Fix migration in P2P mode
6470 * Forbid newline character in names of some libvirt objects
6472 * Fix compilation on macOS
6541 v1.2.21 (2015-11-04)
6542 ====================
6547 v1.2.20 (2015-10-02)
6548 ====================
6553 v1.2.19 (2015-09-02)
6554 ====================
6559 v1.2.18 (2015-08-03)
6560 ====================
6565 v1.2.17 (2015-07-02)
6566 ====================
6571 v1.2.16 (2015-06-01)
6572 ====================
6577 v1.2.15 (2015-05-04)
6578 ====================
6583 v1.2.14 (2015-04-02)
6584 ====================
6589 v1.2.13 (2015-03-02)
6590 ====================
6595 v1.2.12 (2015-01-27)
6596 ====================
6601 v1.2.11 (2014-12-13)
6602 ====================
6607 v1.2.10 (2014-11-03)
6608 ====================
6745 v0.10.2 (2012-09-24)
6746 ====================
6751 v0.10.1 (2012-08-31)
6752 ====================
6757 v0.10.0 (2012-08-29)
6758 ====================
6763 v0.9.13 (2012-08-02)
6764 ====================
6769 v0.9.12 (2012-05-14)
6770 ====================
6775 v0.9.11 (2012-04-03)
6776 ====================
6781 v0.9.10 (2012-02-13)
6782 ====================
7087 v0.1.11 (2007-01-22)
7088 ====================
7093 v0.1.10 (2006-12-20)
7094 ====================
7195 .. _git log: https://gitlab.com/libvirt/libvirt/-/commits/master