[libvirt PATCH] qemu: end the agent job in qemuDomainSetTimeAgent

This function grabs an agent job but ends a monitor job. End the agent job instead. https://bugzilla.redhat.com/show_bug.cgi?id=1792723 Signed-off-by: Ján Tomko Reported-by: Dan Zheng Fixes: e005c95f56fee9ed780be7f8db103d690bd34cbd --- src/qemu/qemu_driver.c | 2 +- 1 file changed, 1

Re: [libvirt PATCH 00/12] virtio failover / vfio auto-plug-on-migrate

I forgot to mention in the cover letter - there is a bug in the QEMU 4.2.0 release that causes the qemu process to crash whenever the vfio device is unplugged, and for some reason when booting a Fedora 31 guest with a failover pair, the guest kernel attempts to unplug the vfio device during

[libvirt PATCH 12/12] conf/qemu: new attribute "useBackupMAC"

Current virtio-net drivers that support the failover feature match up the virtio backup device with its corresponding hostdev device by looking for an interface with a matching MAC address. Since libvirt will assign a different random MAC address to each interface that isn't given an explicit MAC

[libvirt PATCH 09/12] qemu: allow migration with assigned PCI hostdev if backupAlias is set

Normally a PCI hostdev can't be migrated, so qemuMigrationSrcIsAllowedHostdev() won't permit it. In the case of a PCI hostdev that has the backupAlias attribute set, QEMU will automatically unplug the device prior to migration, and re-plug a corresponding device on the destination. This patch

[libvirt PATCH 08/12] conf: add backupAlias attribute to driver subelement

For the subelement (including the backupAlias attribute) is parsed directly into the hostdev child object (virDomaniHostdevDef) of the interface (using virDomainHostdevDefParseXMLSubsys()). But for where the network is a pool of hostdevs, the hostdev object doesn't exist until the network port

[libvirt PATCH 11/12] docs: document virtio failover / QEMU auto-plug of hostdev during migration

The information in formatdomain.html seems too detailed, but it also didn't seem right to put that information in a wiki page before the patches are even pushed... Signed-off-by: Laine Stump --- docs/formatdomain.html.in | 70 +++ docs/news.xml |

[libvirt PATCH 05/12] qemu: add capabilities flag for failover feature

Presence of the virtio-net-pci option called "failover" indicates support in a qemu binary of a simplistic bonding of a virtio-net device with another PCI device. This feature allows migration of guests that have a network device assigned to a guest with VFIO, by creating a network bond device in

[libvirt PATCH 10/12] qemu: add wait-unplug to qemu migration status enum

Aside from eliminating itinerant error (effectively just warning, since the migration continues) messages due to an unrecognized response from qemu, this isn't even necessary - the migration proceeds successfully to completion anyway. (I'm not sure where to see this status reported in the API

[libvirt PATCH 04/12] conf: refactor 's subelement parse/format

Until now the subelement of was only parsed/used/formatted if the device model was "virtio". Making the driver member of the data object a struct instead of a union and refactoring the parsing/formatting code in this separate patch will make it easier to see exactly what is being added when an

[libvirt PATCH 03/12] conf: rename two static functions

Adding Driver to the names makes them better fit their purpose. Signed-off-by: Laine Stump --- src/conf/domain_conf.c | 12 ++-- 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 69e3b7be9a..58f4a1c14b 100644 ---

[libvirt PATCH 02/12] conf: change virDomainVirtioNet*Format() to return void

All three of these functions could only return 0 anyway, so just get rid of all the extra red tape. Signed-off-by: Laine Stump --- src/conf/domain_conf.c | 20 ++-- 1 file changed, 6 insertions(+), 14 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c

[libvirt PATCH 00/12] virtio failover / vfio auto-plug-on-migrate

https://bugzilla.redhat.com/1693587 'QEMU 4.2.0 and later, combined with a sufficiently recent guest virtio-net driver, supports setting up a simple network bond device comprised of one virtio emulated NIC and one hostdev NIC (which must be an SRIOV VF). The allure of this setup is that the bond

[libvirt PATCH 06/12] conf: add failover attribute to subelement of

This attribute is only used for virtio-net devices, so it is stored in the virtio part of the anonymous union in virDomainNetDef::driver. An example of the new option: The corresponding qemu commandline option can only be set if the qemu binary supports it, so

[libvirt PATCH 01/12] conf: refactor hostdev driver subelement format for simpler additions

This reorganizes with no functional change, to make it easier to see what is added in an upcoming patch. (Yes, the "backend != DEFAULT" clause is duplicated in the nested if - the first is to check if *any* attributes of are present, the second is checking for this specific attribute; this will

[libvirt PATCH 07/12] qemu: add backupAlias attribute to subelement of hostdev devices

The new attribute backupAlias gives the alias name ("id" in QEMU parlance) of the virtio-net device that will be paired in the guest with a vfio-pci device for "failover" when the vfio-pci device is unplugged (e.g. during migration). This patch adds backupAlias for (which has its stored in the

[PATCH ocaml 2/3] String_val returns const char * in OCaml 4.10.

This should be compatible with earlier versions of OCaml too since we are just assigning a char * to a const char *. --- libvirt/generator.pl| 14 +++--- libvirt/libvirt_c_oneoffs.c | 6 +++--- 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/libvirt/generator.pl

[PATCH ocaml 3/3] Don't try to memcpy into a String_val.

In OCaml 4.10 String_val returns const char *, so we cannot use it as the destination for memcpy. Use Bytes_val instead. --- libvirt/generator.pl| 2 +- libvirt/libvirt_c_oneoffs.c | 8 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/libvirt/generator.pl

[PATCH ocaml 0/3] Small fixes for OCaml 4.10.

Basically OCaml 4.10 is much more strict about the difference between string (immutable) and bytes (mutable). Rich.

[PATCH ocaml 1/3] block_peek, memory_peek: Use bytes for return buffer.

Strings are immutable in modern OCaml. --- libvirt/libvirt.ml | 4 ++-- libvirt/libvirt.mli | 4 ++-- libvirt/libvirt_c_oneoffs.c | 4 ++-- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/libvirt/libvirt.ml b/libvirt/libvirt.ml index 7f9d0e4..bdb9460 100644 ---