Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Add dbusVMStates to keep a list of dbus-vmstate objects needed for
migration. They are populated on the command line during start or
qemuDBusVMStateAdd/Remove() will hotplug them as needed.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Add a generic way to run a command through the security management.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

pid filenames (from swtpm and other helpers from this series) are
based on VM shortname, which is derived from VM id. If the id is reset
to early, the state filenames will not be found.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

This object is being proposed to qemu upstream "Add dbus-vmstate
object". It handles data migration of external processes.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Datagram socket is available since qemu 4.0, commit
fdec16e3c2a614e2861f3086b05d444b5d8c3406 ("net/socket: learn to talk
with a unix dgram socket").

Required for slirp-helper communication.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Once QEMU is started, the qemuDomainLogContext is owned by it, and can
no longer be used from libvirt. Instead, use
qemuDomainLogAppendMessage() which will redirect the log.

This is not strictly necessary for swtpm, but the following patches
are going to reuse qemuExtDeviceLogCommand().
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

dbus_message_new() does not construct correct replies by itself, it is
recommended to use dbus_message_new_method_return() instead.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

It is failing, because it ends up being parsed with version='default'
and expects '1.2' instead.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Consistent code style across editors.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Implicitly the query depth is limited by the length of the QAPI schema
query, but 'alternate' and 'array' QAPI meta-types don't consume a part
of the query string thus a loop on such types would get our traversal
code stuck in an infinite loop. Prevent this from happening by limiting
the nesting depth to 1000.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

After parsing a video device with a model type of
VIR_DOMAIN_VIDEO_TYPE_NONE, all device info is cleared (see
virDomainDefPostParseVideo()) in order to avoid formatting any
auto-generated values for the XML. Subsequently, however, an alias is
generated for the video device (e.g. 'video0'), which results in an
alias property being formatted in the XML output anyway. This creates
confusion if the user has explicitly provided an alias for the video
device since the alias will change.

To avoid this, don't clear the user-defined alias for video devices of
type "none".

https://bugzilla.redhat.com/show_bug.cgi?id=1720612Signed-off-by: NJonathon Jongsma <jjongsma@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

When we set cpu.max period we need to parse the cpu.max file first as
it contains both quota and period values separated by space.  When only
a single number is written to that file it will set quota.  However,
in order to change period we need to write both values.

The code was prepared for that but mistakenly used new line to end the
string with the first value.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1749227Signed-off-by: NPavel Hrdina <phrdina@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

When blockdev is used we always should use the blockdev mode for
non-shared storage migration.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>

Use VIR_AUTOUNREF and remove the cleanup label.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>

Use VIR_AUTOFREE and remove the cleanup label.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>

Remove the cleanup label as it's empty.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NEric Blake <eblake@redhat.com>

Remove libvirt's support check for the target of an external snapshot to
the blockdev code or qemu. This will potentially require a more complex
cleanup but removes a level of hardcoded feature checks.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Use the code for creating or attaching new storage source in the
snapshot code and switch to 'blockdev-snapshot' for creating the
snapshot itself.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

With blockdev we'll be able to support protocols which are not supported
by the storage backends in libvirt. This means that we have to be able
to skip the creation and relative storage path reading if it's not
supported. This will make it impossible to use relative backing for
network protocols but that would be almost insane anyways.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

After we always assume support for the 'transaction' command
(c358adc5) and follow-up cleanups
qemuDomainSnapshotCreateSingleDiskActive lost its value. Move the code
into appropriate helpers and remove the function.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Fix and unify the naming of external snapshot preparation functions.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Make qemuDomainSnapshotDiskDataCleanup cleanup section friendly by
moving the error preservation code inside it.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

We changed to always saving the status and config XMLs to simplify
code. After a few more refactors it's now possible to move it to the
appropriate place and save the XMLs only on success again.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

When we take a snapshot we must properly remove our locking
infrastructure locks. This was broken by commit 3817fa10 which
attempted to properly track the readonly state for the image as the
locking code was executed after this change. Since we forced the image
which was locked as read-write to read-only prior to unlocking it the
write lock was not dropped.

Fix it by moving the locking code prior to modifying the readonly flag.

https://bugzilla.redhat.com/show_bug.cgi?id=1745618Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

The code preparing data for creating/attaching the target image of block
copy didn't use the correct reference to the existing backing chain in
case when the copy should inherit it. This meant that qemu actually
opened a second copy of the chain and operated on that.

This would de-sync qemu from libvirt's view of node names. Luckily this
is only hypothetical at this point since it happens only when -blockdev
is enabled.

Fix it by passing 'mirrorBacking' which has the proper data as the
backing store when calling
qemuBuildStorageSourceChainAttachPrepareBlockdevTop.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

In some cases we'll need to pass in a backing store which is not
recorded as the backing store of @src. Export backingStore as variable
and fix all callers to pass in the backing store. No semantic changes
for now.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Pass backing store as an argument rather than extracting it locally and
fix the callers.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Extract the loop and supporting infrastructure to the caller as only one
of the two callers actually cares about looping and rename the helper to
qemuBuildStorageSourceChainAttachPrepareBlockdevOne.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Pass in backing store explicitly to qemuBlockStorageSourceGetBlockdevProps
and fix the callers.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Move all bits of the formatting of the 'backing' attribute to a single
condition and make it use a single extracted copy of the backing store.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Since libvirt stores the backing chain into the XML in a nested way it
is the prime possibility to hit libxml2's parsing limit of 256 layers.

Introduce code which will crawl the backing chain and verify that it's
not too deep. The maximum nesting is set to 200 layers so that there's
still some space left for additional properties or nesting into snapshot
XMLs.

The check is applied to all disk use cases (starting, hotplug, media
change) as well as block copy which changes image and snapshots.

We simply report an error and refuse the operation.

Without this check a restart of libvirtd would result in the status XML
failing to be parsed and thus losing the VM.

https://bugzilla.redhat.com/show_bug.cgi?id=1524278Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Use VIR_AUTOUNREF and get rid of the cleanup label.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

With blockdev we must issue the block_set_io_throttle QMP command to
setup disk throttling as we currently can't do it with the 'throttle'
layer.

Unfortunately there's nothing we can do if it fails.

https://bugzilla.redhat.com/show_bug.cgi?id=1733163Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Get rid of the last manually freed var.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Switch to using VIR_AUTOFREE and remove the cleanup label.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

When virtlogd is used to capture QEMU's stdout, qemuDomainObjTaint would
always fail to write the message to the log file when QEMU is already
running (i.e., outside qemuProcessLaunch). This can happen during device
hotplug or by sending a custom QEMU guest agent command:

    warning : qemuDomainObjTaint:8757 : Domain id=9 name='blaf'
        uuid=9cfa4e37-2930-405b-bcb4-faac1829dad8 is tainted:
        custom-ga-command
    error : virLogHandlerDomainOpenLogFile:388 : Cannot open log file:
        '/var/log/libvirt/qemu/blaf.log': Device or resource busy
    error : virNetClientProgramDispatchError:172 : Cannot open log file:
        '/var/log/libvirt/qemu/blaf.log': Device or resource busy

The fix is easy, we just need to use the right API for appending a
message to QEMU log file instead of creating a new log context.
Signed-off-by: NJiri Denemark <jdenemar@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

The attribute is named 'secret', not 'uuid'
Signed-off-by: NCole Robinson <crobinso@redhat.com>

Reviewed-by: NLaine Stump <laine@laine.org>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>