When hotplugging a chardev, Libvirt opens corresponding
file/binds to a socket/does whatever necessary to obtain an FD
that is later passed to QEMU. However, due to wrong placement of
the function that does all of this
(qemuProcessPrepareHostBackendChardevHotplug()) it may happen
that a file is set seclabel on, only to be unlink()-ed and
created again (the former is done by
qemuSecuritySetChardevLabel(), the latter by aforementioned
function). The unlink()-ing is done for UNIX sockets with
mode='bind' and happens inside qemuOpenChrChardevUNIXSocket().
However, these steps can be swapped simply.
Fixes: ad81aa8ad0
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Kristina Hanicova <khanicov@redhat.com>
When hotplugging a chardev, Libvirt opens corresponding
file/binds to a socket/does whatever necessary to obtain an FD
that is later passed to QEMU. However, if something fails after
the FDs were transferred to QEMU and before chardev is actually
added via monitor, these FDs are never closed in QEMU. This is
rather suboptimal.
Fixes: 15bdced9b3
Fixes: ad81aa8ad0
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Kristina Hanicova <khanicov@redhat.com>
Commit 6262752460 added the acquiring of a job, but it is not always
VIR_ASYNC_JOB_MIGRATION_OUT, so the code fails when doing save or anything else.
Correct the async job by passing it from the caller as another parameter.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
libvirt currently will block migration for any vfio-assigned device
unless it is a network device that is associated with a virtio-net
failover device (ie. if the hostdev object has a teaming->type ==
VIR_DOMAIN_NET_TEAMING_TYPE_TRANSIENT).
In the future there will be other vfio devices that can be migrated,
so we don't want to rely on this hardcoded block. QEMU 6.0+ will
anyway inform us of any devices that will block migration (as a part
of qemuDomainGetMigrationBlockers()), so we only need to do the
hardcoded check in the case of old QEMU that can't provide that
information.
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
The new code that queries QEMU about migration blockers was put at the
top of qemuMigrationSrcIsAllowed(), but that function can also be
called in the case of offline migration (ie when the domain is
inactive / QEMU isn't running). This check should have been put inside
the "if (!(flags & VIR_MIGRATE_OFFLINE))" conditional, so let's move
it there.
Fixes: 156e99f686
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
The code is run with an async job and thus needs to make sure a nested
job is acquired before entering the monitor.
While touching the code in qemuMigrationSrcIsAllowed I also fixed the
grammar which was accidentally broken by v8.5.0-140-g2103807e33.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This patch moves qemuDomainObjInitJob() as virDomainObjInitJob()
into hypervisor in order to be used by other drivers as well.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
We have qemuCgroupAllowDevicePath() which sets up devices
controller for just one path. And if we have more paths we have
to call it in a loop. So far, we have just one such place, but
soon we'll have another one (for SGX memory). Separate the loop
into its own function so that it can be reused.
And while at it, move setting the default set of devices as the
first thing, right after all devices are disallowed.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Inside of the qemuSetupDevicesCgroup() there's @deviceACL
variable, which points to a string list of devices that are
allowed in devices controller by default. This list can either
come from qemu.conf (cfg->cgroupDeviceACL) or from a builtin
@defaultDeviceACL. However, a multiline ternary operator is used
when setting the variable which is against our coding style.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
vDPA devices will be migratable soon, so we shouldn't unconditionally
block migration of any domain with a vDPA device. Instead, we should
rely on QEMU to make the decision when that info is available from the
query-migrate QMP command (QEMU versions too old to have that info in
the results of query-migrate don't support migration of vDPA devices,
so in that case we will continue to unconditionally block migration).
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Laine Stump <laine@redhat.com>
Since QEMU 6.0, if QEMU knows that a migration would fail,
'query-migrate' will return an array of error strings describing the
migration blockers. This can be used to check whether there are any
devices/conditions blocking migration.
This patch adds a call to this query at the top of
qemuMigrationSrcIsAllowed().
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Laine Stump <laine@redhat.com>
Since QEMU 6.0, if migration is blocked for some reason,
'query-migrate' will return an array of error strings describing the
migration blockers. This can be used to check whether there are any
devices, or other conditions, that would cause migration to fail.
This patch adds a function that sends this query via a QMP command and
returns the resulting array of reasons. qemuMigrationSrcIsAllowed()
will be able to use the new function to ask QEMU for migration
blockers, instead of the hardcoded guesses that libvirt currently has.
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Laine Stump <laine@redhat.com>
since qemu 6.0, if migration is blocked for some reason, 'query-migrate'
will return an array of error strings describing the migration blockers.
This can be used to check whether there are any devices blocking
migration, etc.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Laine Stump <laine@redhat.com>
QEMU supports hotplug of a cdrom device with USB or SCSI bus. Just
unblock these devices in qemuDomainAttachDeviceDiskLiveInternal() and
qemuDomainDetachPrepDisk().
Fixes: https://gitlab.com/libvirt/libvirt/-/issues/261
Signed-off-by: minglei.liu <minglei.liu@smartx.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This patch alters members of virDomainObjPrivateJobCallbacks to
make the code more consistent.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This patch moves qemuDomainJobObj into hypervisor/ as generalized
virDomainJobObj along with generalized private job callbacks as
virDomainObjPrivateJobCallbacks.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This patch removes variable 'async', which is used only once, and
replaces it with direct comparison with an enum member.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
When formatting qemuCaps XML, the <cpudata/> element is
misaligned. This is because it contains multiple lines and
virBufferAsprintf() does not expect that. Switch to
virBufferAddStr() which does.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
The G_GNUC_NO_INLINE macro will eventually be marked as
deprecated [1] and we are recommended to use G_NO_INLINE instead.
Do the switch now, rather than waiting for compile time warning
to occur.
1: 15cd0f0461
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
These wrapper functions were used to adapt the virObjectUnref() function
signature for different callbacks. But in commit 0d184072, the
virObjectUnref() function was changed to return a void instead of a
bool, so these adapters are no longer necessary.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
The getters/setters for individual properties of migration
speed/downtime/cache size are unused once we switched to setting them
purely via migration parameters. Remove the unused helpers.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The 'xbzrle-cache-size' parameter was added in qemu-2.11 thus all
supported qemu versions now use the new code path.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The 'downtime-limit' field of 'migrate-set-parameters' was introduced in
qemu-2.8, thus all qemu versions supported by libvirt use the new code.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The 'max-bandwidth' field was added as argument of
'migrate-set-parameters' in qemu-2.8, thus all qemu version supported by
libvirt already use the new code path.
This patch assumes the presence and removes the legacy code paths.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
QEMU offers two attributes for handling reset requests of an USB
host device: guest-reset and guest-resets-all. When combined they
act as follows:
1) guest-reset=false
The guest is not allowed to reset the physical USB device.
2) guest-reset=true,guest-resets-all=false
The guest is allowed to reset the device when it is not yet
initialized (aka no USB bus address assigned). Usually this results
in one guest reset being allowed. This is the default behavior.
3) guest-reset=true,guest-resets-all=true
The guest is allowed to reset the device as it pleases.
Now, there's a clear 1:1 mapping with our representation of
guestReset, so generating cmd line is trivial.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Now that we have a capability, validate that the QEMU we are
talking to has everything we need for guestReset.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We will need two attributes of usb-host device to set:
guest-reset and guest-resets-all. The former was introduced in
QEMU v4.0.0-rc0~56^2 and the other in v4.2.0-rc1~9^2. Hence,
track the latter only as it's only starting from that commit when
QEMU has both attributes.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
It was always possible to modify the inactive XML, because
VIR_DOMAIN_AFFECT_CURRENT (= 0) is accepted implicitly. But now
that the logic when changing both config and live XMLs is more
robust we can accept VIR_DOMAIN_AFFECT_CONFIG flag too.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
There are three APIs that allow changing IOThreads:
virDomainAddIOThread()
virDomainDelIOThread()
virDomainSetIOThreadParams()
In case of QEMU driver these are handled by
qemuDomainChgIOThread() which attempts to be versatile enough to
work on both inactive and live domain definitions at the same
time. However, it's a bit clumsy - when a change to live
definition succeeds but fails in inactive definition then there's
no rollback. And somewhat rightfully so - changes to live
definition are in general harder to roll back. Therefore, do what
we do elsewhere (qemuDomainAttachDeviceLiveAndConfig(),
qemuDomainDetachDeviceAliasLiveAndConfig(), ...):
1) do the change to inactive XML first,
2) in fact, do the change to a copy of inactive XML,
3) swap inactive XML and its copy only after everything
succeeded.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
When virDomainSetIOThreadParams() API is called, well its QEMU
impl: qemuDomainSetIOThreadParams() then typed params are parsed
by qemuDomainIOThreadParseParams() into this
qemuMonitorIOThreadInfo struct. In the struct we have a <int,
bool> pair for every IOThread attribute we can tune through
monitor. The struct is then passed to
qemuMonitorJSONSetIOThread() which looks at the bool and if set
then the corresponding attribute is set to given value. Each
attribute is thus changed in a separate call. While this works
for attributes independent of each other ("poll-max-ns",
"poll-grow", "poll-shrink"), it does not always work for the
other attributes ("thread-pool-min" and "thread-pool-max").
The limitation here is that the lower boundary (minimum) has to
be lower (or equal to) the upper boundary (maximum) at all times.
This means, that in some cases we might need to set attributes in
reversed order to meet the constraint.
Resolves: https://gitlab.com/libvirt/libvirt/-/issues/339
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
The domain post parse functions currently live in domain_conf.c
which thus grows always larger. Mimic what we've done for the
validation code and move the post parse code into a separate
file: domain_postparse.c.
I've started by moving every function with PostParse in its name
into the new file and then compile hunting for helper functions
only to move them as well.
In the end, I've moved virDomainDefPostParse symbol in
libvirt_private.syms into a new section. And while
virDomainDeviceDefPostParseOne() is made 'public' in
domain_postparse.h too, I'm not exporting it because it has no
caller outside src/conf/ and it's unlikely it ever will.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
There are two places where the @model member of
_virDomainIOMMUDef struct is typecasted to virDomainIOMMUModel
which is completely unnecessary because the struct already
defines the member of that type.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
No sane firmware build will fail this check, but just to be on
the safe side let's check anyway.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Currently, a firmware configuration such as
<os firmware='efi'>
<firmware>
<feature enabled='yes' name='enrolled-keys'/>
</firmware>
</os>
will correctly pick a firmware that implements the Secure Boot
feature and initialize the NVRAM file so that it contains the
keys necessary to enforce the signing requirements. However, the
lack of a
<loader secure='yes'/>
element makes it possible for pflash writes to happen outside
of SMM mode. This means that the authenticated UEFI variables
where the keys are stored could potentially be overwritten by
malicious code running in the guest, thus making it possible to
circumvent Secure Boot.
To prevent that from happening, automatically turn on the
loader.secure feature whenever a firmware that implements Secure
Boot is chosen by the firmware autoselection logic. This is
identical to the way we already automatically enable SMM in such
a scenario.
Note that, while this is technically a guest-visible change, it
will not affect migration of existings VMs and will not prevent
legitimate guest code from running.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
When resuming post-copy migration users may want to limit the bandwidth
used by the migration and use a value that is different from the one
specified when the migration was originally started.
Resolves: https://gitlab.com/libvirt/libvirt/-/issues/333
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
So the we can apply selected migration parameters even when resuming
post-copy migration.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
We will need to annotate individual parameters a bit more than just
noting their type. Let's introduce qemuMigrationParamInfo replacing
simple qemuMigrationParamTypes with an array of structs.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The flags will later be used to determine which parameters should
actually be applied.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
My original commit v8.4.0-288-gf01fc4d119 accidentally forgot to fix
both instances of the same problem. While it fixed the destination side
of migration, the source one remained broken.
However, that commit was also wrong in saying the issue could have
caused unlimited memory locking to be allowed for QEMU when RDMA
migration was used. It could not, because the code would refuse to even
think about starting RDMA migration if hard_limit was not set. But
avoiding the "mem.hard_limit > 0" check is useful anyway.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Introduced in v8.4.0-rc1~183 but the first real problem
introduced in v8.4.0-rc1~170, there's a
qemuBuildInterfaceConnect() call inside of
qemuDomainAttachNetDevice(). If the former fails, then the
function is immediately returned from instead of jumping onto the
cleanup label. This is crucial, because at this point the domain
definition contains 'borrowed' net definition, which is then
freed, since an error was met. The domain definition is then left
with a dangling pointer which leads to all sorts of different
crashes.
Fixes: 29d022b1eb
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2102009
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Our documentation says RDMA migration requires hard_limit to be set so
that we know how big memory locking limit should be set for the domain
during migration. But since commit v1.2.13-71-gcf521fc8ba (which changed
the default hard_limit value from 0 to
VIR_DOMAIN_MEMORY_PARAM_UNLIMITED) we were actually setting memlock
limit to unlimited if hard_limit was not set.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
For RDMA migration we update memory locking limit, but never set it back
once migration finishes (on the destination host) or aborts (on the
source host).
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This helper will not try to set the limit if it is already big enough,
which may be useful when libvirt daemon is running in a containerized
environment and is not allowed to change memory locking limit.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
qemuDomainAdjustMaxMemLock combined computing the desired limit with
applying it. This patch separates the code to apply a memory locking
limit to a new qemuDomainSetMaxMemLock helper for better reusability.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
