QEMU 6.1 will replace the virgl property of virtio-vga device to
virtio-vga-gl device. Adapt to that update.
Resolves: https://gitlab.com/libvirt/libvirt/-/issues/167
Signed-off-by: Han Han <hhan@redhat.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
QEMU 6.1 will add virtio-gpu-gl-pci device to replace the virgl property
of virtio-gpu-pci device. Adapt to that change.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1967356
Signed-off-by: Han Han <hhan@redhat.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The devices virtio-gpu-gl-pci and virtio-vga-gl, aimed to replace the
virgl property, are valid for 3d accerlation as well.
Signed-off-by: Han Han <hhan@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This flag will be used for the device virtio-gpu-gl-pci which is introduced
since QEMU 6.1.
Signed-off-by: Han Han <hhan@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The "spawnDaemon" and "binary" parameters are co-dependant, with the
latter non-NULL, if-and-only-if the former is true. Getting rid of the
"spawnDaemon" parameter simplifies life for the callers and eliminates
an error checking scenario.
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
libacl is Linux-only, so we don't need to explicitly check for
either the target platform or header availability, and we can
simply rely on cc.find_library() instead. The corresponding
preprocessor define is renamed to more accurately reflect the
nature of the check.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
QEMU_DOMAIN_DISK_PRIVATE(disk)->transientOverlayCreated flag
gets true unexpectedly on qemuProcessSetupDisksTransientSnapshot() when
the disk has <transient shareBacking='yes'> option.
The flag should be enabled on qemuDomainAttachDiskGeneric() after the
overlay setup is completed.
Skip enabling transientOverlayCreated for the disk here.
Fixes: 75871da0ecb8b552f9e304d0f83e216839bbf82d
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
When <transient shareBacking='yes'> is set to a disk and the overlay
disk already exists because of something abnormal, libvirt is terminated
by Segmentation fault.
# virsh start Test0
error: Disconnected from qemu:///system due to end of file
error: Failed to start domain 'Test0'
error: End of file while reading data: Input/output error
Add NULL check for snapdiskdef so that the rollback can work correctly.
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Fixes: 2e94002d2ace4e4a6dbfc13a84fdab28f22c5c4a
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Ján Tomko <jtomko@redhat.com>
and re-adjust if the hotplug fails.
This fixes a bug found during testing of
https://bugzilla.redhat.com/1939776, which was supposed to be resolved
by commit 98e22ff749, but failed to account for the case of device
hotplug.
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
An upcoming patch will be checking if the addition of a new net device
requires adjusting the domain locked memory limit, which must be done
prior to sending the command to qemu to add the new device. But
qemuDomainAdjustMaxMemLock() checks all (and only) the devices that
are currently in the domain definition, and currently we are adding
new net devices to the domain definition only at the very end of the
hotplug operation, after qemu has already executed the device_add
command.
In order for the upcoming patch to work, this patch changes
qemuDomainAttachNetDevice() to add the device to the domain nets list
at an earlier time. It can't be added until after PCI address and
alias name have been determined (because both of those examine
existing devices in the domain to figure out a unique value for the
new device), but must be done before making the qemu monitor call.
Since the device has been added to the list earlier, we need to
potentially remove it on failure. This is done by replacing the
existing call to virDomainNetRemoveHostdev() (which checks if this is
a hostdev net device, and if so removes it from the hostdevs list,
since it could have already been added to that list) with a call to
the new virDomainNetRemoveByObj(), which looks for the device on both
nets and hostdevs lists, and removes it where it finds it.
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
We have many places where the earliest error returns from a function
skip any cleanup label at the bottom (the assumption being that it is
so early in the function that there isn't yet anything that needs to
be explicitly undone on failure). But in general it is a bad sign if
there are any direct "return" statements in a function at any time
after there has been a "goto cleanup" - that indicates someone thought
that an earlier point in the code had done something needing cleanup,
so we shouldn't be skipping it.
There were two occurences of a "return -1" after "goto cleanup" in
qemuDomainAttachDeviceNet(). The first of these has been around for a
very long time (since 2013) and my assumption is that the earlier
"goto cleanup" didn't exist at that time (so it was proper), and when
the code further up in the function was added, the this return -1 was
missed. The second was added during a mass change to check the return
from qemuInterfacePrepareSlirp() in several places (commit
99a1cfc43889c6d); in this case it was erroneous from the start.
Change both of these "return -1"s to "goto cleanup". Since we already
have code paths earlier in the function that goto cleanup, this should
not cause any new problem.
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Implement this behaviour by skipping the disks on traditional
commandline and hotplug them before resuming CPUs. That allows to use
the support for hotplugging of transient disks which inherently allows
sharing of the backing image as we open it read-only.
This commit implements the validation code to allow it only with buses
supporting hotplug and the hotplug code while starting up the VM.
When we have such disk we need to issue a system-reset so that firmware
tables are regenerated to allow booting from such device.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The qemuDomainAttachDiskGeneric will also be used on startup for
transient disks which share the overlay. The VM startup code passes the
asyncJob around so we need to pass it into qemuDomainAttachDiskGeneric.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Add code which creates the transient overlay after hotplugging the disk
backend before attaching the disk frontend.
The state of the topmost image is modified to be already read-only to
prevent the need to open the image in read-write mode.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
In preparation for hotplug of <transient> disks we'll need to track
whether the overlay file was created individually per-disk.
Add 'transientOverlayCreated' to 'struct _qemuDomainDiskPrivate' and
remove 'inhibitDiskTransientDelete' from 'qemuDomainObjPrivate' and
adjust the code for the change.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Split up the monitor contexts to attach the backend of the disk and the
frontend device in preparation for hotplugging transient disks where
we'll need to add the code for adding the transient overlay between
these two steps.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Modify the rollback section to use its own monitor context so that we
can later split up the hotplug into multiple steps and move the
detachment of the extension device into the rollback section rather than
doing it inline.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Similarly to previous refactors we want to move all hotplug related
setup which isn't strictly relevant to attaching the disk into
qemuDomainAttachDeviceDiskLiveInternal.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Remove the 'ret' variable and 'cleanup' label in favor of directly
returning the value since we don't have anything under the 'cleanup:'
label.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Remove two empty lines.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Move the auditing entry and insertion into the disk definition from the
function which deals with qemu to 'qemuDomainAttachDeviceDiskLiveInternal'
which deals with the hotplug related specifics.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
qemuDomainAttachDeviceDiskLiveInternal already sets up certain pieces of
the disk definition so it's better suited to move the setup of the
virStorageSource structs, granting access to the storage and allocation
of the alias from qemuDomainAttachDiskGeneric which will be just
handling the qemu interaction.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
We can call it in one place as all per-device-type subcases use the same
code.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Move the validation of the SCSI device address and the attachment of the
controller into qemuDomainAttachDeviceDiskLiveInternal as there's no
specific need for a special helper.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Move the specific device setup and address reservation code into the
main hotplug helper as it's just one extra function call.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Move the specific device setup and address reservation code into the
main hotplug helper as it's just one extra function call.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Unify the handling of the copy-on-read filter by changing the handling
to use qemuBlockStorageSourceChainData.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Fill in the required fields in qemuBlockStorageSourceChainData to handle
the hotplug so that we can simplify the cleanup code.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
qemuBlockStorageSourceChainData encapsulates the backend of the disk for
startup and hotplug operations. Add the handling for the copy-on-read
filter so that the hotplug code doesn't need to have separate cleanup.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Pre-extending the disk array size is pointless nowadays since we've
switched to memory APIs which don't return failure.
Switch all uses of reallocation of the array followed by
'virDomainDiskInsertPreAlloced' with direct virDomainDiskInsert.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The "machine-loadparm-multiple-disks-nets-s390" case now requires the
QEMU_CAPS_CCW feature to pass validation.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
We can skip the formatting of the bootindex for floppies directly at the
place where it's being formatted.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The logic assigning the bootindices from the legacy boot order
configuration was spread through the command line formatters for the
disk device and for the floppy controller.
This patch adds 'effectiveBootindex' property to the disk private data
which holds the calculated boot index and moves the logic of determining
the boot index into 'qemuProcessPrepareDomainDiskBootorder' called from
'qemuProcessPrepareDomainStorage'.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Later patches will implement sharing of the backing file, so we'll need
to be able to discriminate the overlays per VM.
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The code deals with the startup of the VM and just uses the snapshot
code to achieve the desired outcome.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
We store the virQEMUDriverConfig object in the context.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Remove all the arguments which are present in qemuSnapshotDiskContext.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The config is used both with the preparation and execution functions, so
we can store it in the context to simplify other helpers.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Rather than filling various parts of the context from arguments pass in
the whole context.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Creating the overlay for the disk is needed when starting a new VM only.
Additionally for now migration with transient disks is forbidden
anyways.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The code will be later reused when adding support for sharing the
backing image of the snapshot.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
None of them are currently needed to pass our upstream CI, most were
either for ancient clang versions or coverity for silencing false
positives.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
They were added mostly randomly and we don't really want to keep working
around of false positives.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This reverts commit <1b22dd6dd44202094e0f78f887cbe790c00e9ebc>.
First of all, the reverted commit is incomplete. It only sets
cpuset.mems in the VM root cgroup when the API is used but there is no
code that would do the same when the VM is started.
Libvirt never places any process into the VM root cgroup directly. All
the supporting processes like slirp-helper or dbus-daemon are placed
into the emulator sub-cgroup and all the QEMU threads are distributed
between emulator, vcpu* and iothread* sub-cgroups. The scenario
described in the reverted commit can happen only if someone manually
adds any process there which we should not care about.
If we would like to set the limit in the VM root cgroup we need to
introduce better logic:
- set both (old and new) numa group in the VM root cgroup
- change the numa group in all sub-cgroups to new value
- finally set only the new value in the VM root cgroup
The simplest fix now is to revert the commit.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The description of the function says that the return value is a
file descriptor on success and negative errno on failure which is
not true. If the 'if' case with check on security labels fails,
the return value is -1 not -errno. The solution is to return
'-EINVAL' instead.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Previously, nvram file was created with user/group owner as
'root', rather than specifications defined in libvirtd.conf. The
solution is to call qemuDomainOpenFile(), which creates file with
defined permissions and qemuSecurityDomainSetPathLabel() to set
security label for created nvram file.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1783255
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
