Paths for external devices (well, so far only vTPM) are not
stored in the status XML. Therefore, we need to regenerate them
after we've been restarted and reconnecting to a running domain.
Otherwise these will remain NULL which may later lead to a NULL
dereference.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2150760
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This function is going to be called outside of qemu_extdevice.c.
Expose it to the rest of the driver.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The path generation phase belongs conceptually into domain
preparation phase and not host preparation. Move
qemuExtDevicesInitPaths() call from qemuExtDevicesPrepareHost()
into qemuExtDevicesPrepareDomain().
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The domain startup process is split into multiple phases. One of
them is preparing the domain (at that point live) XML, private
data, various paths, etc - see qemuProcessPrepareDomain(); the
other prepares the host - see qemuProcessPrepareHost(). It's
obvious that the domain XML preparation function must be called
before the host preparation function (e.g. the host preparation
might try to create a file which path is generated in the domain
preparation phase). Nevertheless, let's document this
expectation.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The function is now unused. Remove it to dissuade anybody from trying to
use it in the future.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'virDomainDeviceDefCopy' formats the definition and parses it back.
Since we already are parsing the XML here, we're better off parsing it
twice and save the formatting step.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'virDomainDeviceDefCopy' formats the definition and parses it back.
Since we already are parsing the XML here, we're better off parsing it
twice and save the formatting step.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Remove the 'cleanup' label and 'ret' variable as we can now directly
return form all cases.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'virDomainDeviceDefCopy' formats the definition and parses it back.
Since we already are parsing the XML here, we're better off parsing it
twice and save the formatting step.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'virDomainDeviceDefCopy' formats the definition and parses it back.
Since we already are parsing the XML here, we're better off parsing it
twice and save the formatting step.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Use of qemuDomainValidateVcpuInfo in the helpers for hotplug and unplug
of vCPUs can lead to spurious errors reported such as:
internal error: qemu didn't report thread id for vcpu 'XX'"
The reason for this is that qemuDomainValidateVcpuInfo validates the
state of all vCPUs against the expected state of vCPUs. If an unplug
operation completed before libvirt was unable to process it yet the
expected state could not reflect the current state.
To avoid spurious errors the qemuDomainHotplugAddVcpu and
qemuDomainRemoveVcpu functions are modified to do localized validation
only for the vCPUs they actually modify.
We also now ensure that the cgroups are modified before bailing out on
error for any vCPUs which passed validation.
Additionally in order for qemuDomainRemoveVcpuAlias to be able to find
the unplugged vCPU we must ensure that qemuDomainRefreshVcpuInfo does
not clear out the alias in case when the vCPU is no longer reported by
qemu.
Co-authored-by: Partha Satapathy <partha.satapathy@oracle.com>
Signed-off-by: Shaleen Bathla <shaleen.bathla@oracle.com>
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Recently, the QEMU driver gained support for migration with TPM
state on a shared volume (e.g. NFS). As a part of that, the
destination side avoids setting seclabels on it to avoid cutting
off the source while it is still using it. Makes sense, except
for a wee bit: the secdriver API does a bit more - it also sets
label on the swtpm log file. And this one definitely needs to be
labeled (it lives under /var/log/swtpm/libvirt/qemu/..., i.e. not
on a shared volume).
Previously, qemuSecurityStartTPMEmulator() took care of that. But
during rework to shared volume migration, the code was changed so
now plain qemuSecurityCommandRun() would be run (i.e. no
relabelling).
But after previous commits, we can now chose whether the TPM
state should be relabelled or just the log file.
Fixes: 2e669ec789231d39e0d5f5f6a201d2a661b8070c
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2130192#c7
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This is basically just a continuation of the previous commit.
Now that the security driver APIs have a boolean flag that
controls setting/restoring seclabel of either both TPM state and
log files, or just the log file, propagate this boolean into
those APIs that start/stop swtpm emulator. For now, just pass
true. The juicy bits are soon to come.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The virSecurityDomainSetTPMLabels() and
virSecurityDomainRestoreTPMLabels() APIs set/restore label on two
files/directories:
1) the TPM state (tpm->data.emulator.storagepath), and
2) the TPM log file (tpm->data.emulator.logfile).
Soon there will be a need to set the label on the log file but
not on the state. Therefore, extend these APIs for a boolean flag
that when set does both, but when unset does only 2).
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Introduce a simple helper fetching a sub-element node by name. This is
meant as a simple replacement for either open-coded versions of this or
use of XPath for this trivial lookup.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Use virJSONValueObjectGetArray + virJSONValueArrayToStringList instead
so that the ofvirJSONValueObjectGetStringArray wrapper can be removed.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
In two instances (qemuMonitorJSONGetStringListProperty,
qemuMonitorJSONGetStringArray) the return value is checked by
qemuMonitorJSONCheckReply and extracted by
virJSONValueObjectGetStringArray.
We can use qemuMonitorJSONGetReply which returns it directly and then
virJSONValueArrayToStringList to convert it without the additional
lookup.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Using 'virJSONValueObjectHasKey' when we want to access the value
afterwards is wasteful. Fetch the JSON value right away.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Rather than checking that the object has the correct key and then
fetching it again use fetch the array first and then use
virJSONValueArrayToStringList to directly convert it.
Additionally we can avoid the conversion if there are no members
simplifying the surrounding logic.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The 'dependencies' field in the return data may be missing in some
cases. Historically 'virJSONValueObjectGetStringArray' didn't report
error in such case, but later refactor (commit 043b50b948ef3c2 ) added
an error in order to use it in other places too.
Unfortunately this results in the error log being spammed with an
irrelevant error in case when qemuAgentGetDisks is invoked on a VM
running windows.
Replace the use of virJSONValueObjectGetStringArray by fetching the
array first and calling virJSONValueArrayToStringList only when we have
an array.
Fixes: 043b50b948ef3c2a4adf5fa32a93ec2589851ac6
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2149752
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Introduce virJSONValueArrayToStringList which does only the conversion
from an array to a stringlist.
This will allow refactoring the callers to be more careful in case when
they want to handle the existance of the member in the parent object
differently.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Use qemuMonitorJSONGetReply and unify the two blocks with the same
condition.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Use qemuMonitorJSONGetReply in cases where qemuMonitorJSONCheckReply
is followed by virJSONValueObjectGet*(reply, "return").
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Replace usage of the following pattern with the new helper:
if (qemuMonitorJSONCheckReply(cmd, reply, VIR_JSON_TYPE_ARRAY) < 0)
return -1;
data = virJSONValueObjectGetArray(reply, "return");
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Replace usage of the following pattern with the new helper:
if (qemuMonitorJSONCheckReply(cmd, reply, VIR_JSON_TYPE_OBJECT) < 0)
return -1;
data = virJSONValueObjectGetObject(reply, "return");
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Rather than simply checking that the 'return' field is of the expected
type we can directly return it as the caller is very likely going to use
it. Extract the code into the new function and add a wrapper to preserve
old functionality.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Fix the type for few internal functions. Externally the APIs were
already limiting 'flags' to 'unsigned int'.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Don't continue with the historical mistake and fix all internal
functions to use a sane type for flags.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Historically our migration APIs declare 'unsigned long flags'. Since
it's baked into our API we can't change that but we can avoid
compatibility problems by preemptively refusing the extra range on
certain arches to prevent future surprise.
Modify the macro to verify that value passed inside 'flags' doesn't
exceed the range of 'unsigned int'.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The virCommandSetSendBuffer() function consumes passed @buffer,
but takes it only as plain pointer. Switch to a double pointer to
make this obvious. This allows us then to drop all
g_steal_pointer() in callers.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
Back in v1.0.3-rc1~235 when I was adding virCommandDoAsyncIO(),
the main event loop was used to poll() on the pipe to the child
process. But this was promptly changed to a separate thread
handling I/O in v1.0.3-rc1~127. However, the corresponding
comment to virCommandDoAsyncIO() still documents the original
state.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
When virCommandSetSendBuffer() is used over a virCommand that is
(or will be) daemonized, then the command must have
VIR_EXEC_ASYNC_IO flag set no later than at virCommandRunAsync()
phase so that the thread that's doing IO is spawned and thus
buffers can be sent to the process.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
Error message reports that the guest has '0' NUMA nodes
configured when trying to attach a memory device to a guest with
no NUMA nodes. This may be a little misleading because '0' can
also be node's id. A more friendly way is to directly report
that the guest has no NUMA nodes.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2142519
Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
QEMU capabilities is the only thing we use from priv so we can just pass
that directly.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
When an internal API takes a vm pointer, it's usually just after the
driver argument.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
The vm object is used inside qemuMigrationCookieParse based on the flags
passed to qemuMigrationCookieParse and the content of the cookie. The
callers should not just blindly guess and pass NULL if they
(incorrectly) think the vm object is not needed. We should always pass
the vm object unless it does not exist yet.
This fixes a bug when statistics of a completed migration reported
"Unknown" operation instead of "Incoming migration" on the destination
host.
https://bugzilla.redhat.com/show_bug.cgi?id=2137298
Fixes: v8.7.0-79-g0150f7a8c1
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The virNodeDeviceGetPCIVPDDynamicCap() function is called from
virNodeDeviceGetPCIDynamicCaps() and therefore has to be a wee
bit more clever about adding VPD capability. Namely, it has to
remove the old one before adding a new one. This is how other
functions called from virNodeDeviceGetPCIDynamicCaps() behave
as well.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2143235
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
As of [1]. libselinux changed the type of context_str() - it now
returns a const string. Follow this change in our code base.
1: dd98fa3227
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Inside of qemuTPMEmulatorBuildCommand() there are two calls to
qemuTPMSetupEncryption() which simply ignore returned error. This
is suboptimal because then we rely on swtpm binary reporting a
generic error (something among invalid command line arguments)
while an error reported by qemuTPMSetupEncryption() is more
specific.
However, since virCommandSetSendBuffer() only sets an error
inside of virCommand structure (the error is then reported in
virCommandRun()), we need to exempt its retval from error
checking. Thus, the signature of qemuTPMSetupEncryption() is
changed a bit so that -1/0 can be returned to indicate error.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
In commit c43718ef67944 I've added a disclaimer that the new stats which
are fetched from qemu and passed directly to the user are not guaranteed
by libvirt. I didn't notice that per-vcpu hypervisor specific stats are
also snuck into the VIR_DOMAIN_STATS_VCPU group along with other
pre-existing stats we do guarantee.
Extend the disclaimer for VIR_DOMAIN_STATS_VCPU too.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Guests are allowed to change their MAC addresses. Subsequently,
we may respond to that with tweaking that part of host side
configuration that depends on it. In this particular case: QoS.
Some parts of QoS are in fact set on corresponding bridge, where
overall view on traffic can be seen. Here, TC filters are used to
place incoming packets into qdiscs. These filters match source
MAC address. Therefore, upon guest changing its MAC address, the
corresponding TC filter needs to be updated too. This is done by
simply removing the old one and instantiating a new one, with new
MAC address.
Now, u32 filters (which we use) use a hash table for matching,
internally. And when deleting the old filter, we used to remove
the hash table (ID = 800::) and let the new filter instantiate
new hash table. This used to work, until kernel release 4.20
(specifically commit v4.20-rc1~27^2~131^2~11 and its friends)
where this practice was turned into error.
But that's okay - we can delete the specific filter we are after
and not touch the hash table at all.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
When setting up QoS for a domain <interface/>, or when reporting
its statistics we may need to swap TX/RX values. This is all
explained in comment to virDomainNetTypeSharesHostView().
However, this function claims that VIR_DOMAIN_NET_TYPE_ETHERNET
also shares the 'host view', meaning the TX/RX values must be
swapped. But that's not true.
An easy reproducer is to start a domain with two <interface/>-s:
one type of network, the other of type ethernet and configure the
same <bandwidth/> for both. Reversed setting can then be observed
(e.g. via tc).
Reported-by: Oleg Vasilev <oleg.vasilev@virtuozzo.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Our RPC calls can be divided into two groups: regular and high
priority. The latter can be then processed by so called high
priority worker threads. This is our way of defeating a
'deadlock' and allowing some RPCs to be processed even when all
(regular) worker threads are stuck. For instance: if all regular
worker threads get stuck when talking to QEMU on monitor, the
virDomainDestroy() can be processed by a high priority worker
thread(s) and thus unstuck those threads.
Now, this is all fine, except if users want to use virsh
non interactively:
virsh destroy $dom
This does a bit more - it needs to open a connection. And that
consists of multiple RPC calls: AUTH_LIST,
CONNECT_SUPPORTS_FEATURE, CONNECT_OPEN, and finally
CONNECT_REGISTER_CLOSE_CALLBACK. All of them are marked as high
priority except the last one. Therefore, virsh just sits there
with a partially open connection.
There's one requirement for high priority calls though: they can
not get stuck. Hopefully, the reason is obvious by now. And
looking into the server side implementation the
CONNECT_REGISTER_CLOSE_CALLBACK processing can't ever get stuck.
The only driver that implements the callback for public API is
Parallels (vz). And that can't block really.
And for virConnectUnregisterCloseCallback() it's the same story.
Therefore, both can be marked as high priority.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2143840
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Use same style in the 'struct option' as:
struct option opt[] = {
{ a, b },
{ a, b },
...
{ a, b },
};
Signed-off-by: Jiang Jiacheng <jiangjiacheng@huawei.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
For the handling of usb we already allow plenty of read access,
but so far /sys/bus/usb/devices only needed read access to the directory
to enumerate the symlinks in there that point to the actual entries via
relative links to ../../../devices/.
But in more recent systemd with updated libraries a program might do
getattr calls on those symlinks. And while symlinks in apparmor usually
do not matter, as it is the effective target of an access that has to be
allowed, here the getattr calls are on the links themselves.
On USB hostdev usage that causes a set of denials like:
apparmor="DENIED" operation="getattr" class="file"
name="/sys/bus/usb/devices/usb1" comm="qemu-system-x86"
requested_mask="r" denied_mask="r" ...
It is safe to read the links, therefore add a rule to allow it to
the block of rules that covers the usb related access.
Signed-off-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
Reviewed-by: Michal Privoznik <mprivozn at redhat.com>
When there is no vIOMMU, vfio devices don't need to lock the entire guest
memory per-device, but they still need to lock the entire guest memory to
share between all vfio devices. This memory accounting is not shared
with vDPA devices, so it should be added to the memlock limit separately.
Commit 8d5704e2 added support for multiple vfio/vdpa devices but
calculated the limits incorrectly when there were both vdpa and vfio
devices and no vIOMMU. In this case, the memory lock limit was not
increased separately for the vfio devices.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2143838
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
Reviewed-by: Laine Stump <laine@redhat.com>
