Convert all code using the 'QOM_CPU_PATH' macro to accept the QOM path
as an argument.
For now the new helper for fetching the path 'qemuProcessGetVCPUQOMPath'
will always return the same hard-coded value.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
As the parent address is part of the mdev nodedev name lets expose the
internally available parent address in the XML.
Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
SmartNIC DPUs may not expose some privileged eswitch operations
to the hypervisor hosts. For example, this happens with Bluefield
devices running in the ECPF (default) mode for security reasons. While
VF MAC address programming is possible via an RTM_SETLINK operation,
trying to set a VLAN ID in the same operation will fail with EPERM.
The equivalent ip link commands below provide an illustration:
1. This works:
sudo ip link set enp130s0f0 vf 2 mac de:ad:be:ef:ca:fe
2. Setting (or clearing) a VLAN fails with EPERM:
sudo ip link set enp130s0f0 vf 2 vlan 0
RTNETLINK answers: Operation not permitted
3. This is what Libvirt attempts to do today (when trying to clear a
VF VLAN at the same time as programming a VF MAC).
sudo ip link set enp130s0f0 vf 2 vlan 0 mac de:ad:be:ef:ca:fe
RTNETLINK answers: Operation not permitted
If setting an explicit VLAN ID results in an EPERM, clearing a VLAN
(setting a VLAN ID to 0) can be handled gracefully by ignoring the
EPERM error with the rationale being that if we cannot set this state
in the first place, we cannot clear it either.
In order to keep explicit clearing of VLAN ID working as it used to
be passing a NULL pointer for VLAN ID is used.
Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
There should be a way to show no intent in programming a VLAN at all
(including clearing it). This allows handling error conditions
differently when VLAN clearing is explicit (vlan id == 0) vs implicit
(vlanid == NULL - try to clear it if possible).
Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This has a benefit of being able to handle error codes for those
operations separately which is useful when drivers allow setting a MAC
address but do not allow setting a VLAN (which is the case with some
SmartNIC DPUs).
Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The XML-to-XML test validates that we don't accidentally copy the
isa-debug <serial> into a <console>.
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
This is a perfectly valid configuration that we need to keep
working, so add test coverage for it.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Currently, memory device (def->mems) part of cmd line is
generated before any controller. In majority of cases it doesn't
matter because neither of memory devices live on a bus that's
created by an exposed controller (e.g. there's no DIMM
controller, at least not exposed). Except for virtio-mem and
virtio-pmem, which do have a PCI address. And if it so happens
that the device goes onto non-default bus (pci.0) starting such
guest fails, because the controller that creates the desired bus
wasn't processed yet. QEMU processes arguments in order.
For instance, if virtio-mem has address with bus='0x01' QEMU
refuses to start with the following message:
Bus 'pci.1' not found
Similarly for virtio-pmem. I've successfully tested migration and
changing the order does not affect migration stream.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2047271
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Inside the testPCIVPDResourceCustomCompareIndex() function we
have two variables @a and @b, both marked as g_autoptr(). Then,
towards the end of the function b->value is freed and set to
a->value. This is to make sure
virPCIVPDResourceCustomCompareIndex() works correctly even if
->value member is the same for both arguments.
Nevertheless, if the function returns anything else than 0 then
the control executes subsequent return statement and since
b->value points to the very same string as a->value a double free
will occur. Avoid this by setting b->value to NULL explicitly,
just like we are already doing for the successful path.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
There are a few places where a variable is VIR_FREE()-d and then
explicitly set to NULL. This is not necessary since VIR_FREE()
does that for us.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
Since its introduction in
commit 907a39e735
Add a test suite for validating SELinux labelling
this function did not return NULL on OOM.
Since we abort on OOM now, switch testSELinuxMungePath to void,
return NULL explicitly on XML parsing failure and remove
the (now pointless) cleanup label.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Reporting hv-* properties properly requires hv to be enabled,
see qemu commit 071ce4b03b.
Signed-off-by: Tim Wiederhake <twiederh@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
There are a some scenarios in which we want to prealloc guest
memory (e.g. when requested in domain XML, when using hugepages,
etc.). With 'regular' <memory/> models (like 'dimm', 'nvdimm' or
'virtio-pmem') or regular guest memory it is corresponding
memory-backend-* object that ends up with .prealloc attribute
set. And that's desired because neither of those devices can
change its size on the fly. However, with virtio-mem model things
are a bit different. While one can set .prealloc attribute on
corresponding memory-backend-* object it doesn't make much sense,
because virtio-mem can inflate/deflate on the fly, i.e. change
how big of a portion of the memory-backend-* object is exposed to
the guest. For instance, from a say 4GiB module only a half can
be exposed to the guest. Therefore, it doesn't make much sense to
preallocate whole 4GiB and keep them allocated. But we still want
the part exposed to the guest preallocated (when conditions
described at the beginning are met).
Having said that, with new enough QEMU the virtio-mem-pci device
gained new attribute ".prealloc" which instructs the device to
talk to the memory backend object and allocate only the requested
portion of memory.
Now, that our algorithm for setting .prealloc was isolated in a
single function, the function can be called when constructing cmd
line for virtio-mem-pci device.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This new capability tracks whether virtio-mem device is capable
of memory preallocation, which is detected by the device having
.prealloc attribute.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We need to use a hardcoded list of capabilities because we don't
yet have proper replies files obtained from QEMU running on actual
macOS machines.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Tested-by: Brad Laue <brad@brad-x.com>
Tested-by: Christophe Fergeau <cfergeau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
The new DO_TEST_MACOS() macro makes it possible to create test
cases that verify the behavior of libvirt on a macOS machine
with HVF support available.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Tested-by: Brad Laue <brad@brad-x.com>
Tested-by: Christophe Fergeau <cfergeau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
This exposes a couple of macOS-specific variants of existing
APIs, which can be used when implementing test programs and
result in HVF support being advertised.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Tested-by: Brad Laue <brad@brad-x.com>
Tested-by: Christophe Fergeau <cfergeau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
This new enumeration provides a way to specify the host OS
that a specific test case expects. The default is Linux, which
has been the implicit host OS until now; when Linux is selected
as the host OS, KVM support is advertised in capabilies data
exposed to test cases.
This commit doesn't result in any functional change, and simply
sets the stage for introducing macOS host OS support later.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Tested-by: Brad Laue <brad@brad-x.com>
Tested-by: Christophe Fergeau <cfergeau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
When trying to attach vhost-user-blk device to virtual machine using
qemu < 4.2 libvirt would mistakenly add a scsi=off parameter, which is
not supported by qemu.
Fixes: https://gitlab.com/libvirt/libvirt/-/issues/265
Signed-off-by: shenjiatong <yshxxsjt715@gmail.com>
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
With qemu versions prior to qemu-5.0 we'll format 'scsi=off' for
virtio-blk disks, but also for vhost-user-blk. This is a bug as it's not
supported.
Add a test case to show that wrong configuration is generated by adding
running 'disk-vhostuser' test case on capabilities from qemu-4.2.
For this to be possible it's required to enable shared memory via NUMA
configuration as old QEMU's don't allow configuration of the default
memory backend. This is achieved by adding a copy of the
'disk-vhostuser' XML with NUMA enabled.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
After previous cleanups, the virDomainNetDefParseXML() function
uses a mixture of virXMLProp*() and the old virXMLPropString() +
virXXXTypeFromString() patterns. Rework it so that virXMLProp*()
is used.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Update existing ppc64 6.2 caps to match what was released in QEMU 6.2.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Changes in all 'ppc64-latest.ags' files were needed due to the
JSONification of command line devices.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Since there's no capability to check now, we can simply move the
formatting of 'max_outputs' earlier.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Both are supported by qemu-2.11 and later, so we don't have to check for
them explicitly.
Note that QXL is supported only on x86_64, thus on other arches only the
capability for 'virtio-gpu' is removed.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Both the QXL video device and 'virtio' video device support
'max_outputs' in all qemu versions libvirt supports. This means we no
longer have to check the QEMU_CAPS_QXL_MAX_OUTPUTS and
QEMU_CAPS_VIRTIO_GPU_MAX_OUTPUTS capabilities.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
It was only used to construct the hash key for the (now removed)
shared devices in the qemu driver.
Remove it and its mocking.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
No kernels supported by upstream libvirt have the feature.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
This device was virtio 1.0-only so adding the (non-)transitional model
did not make sense and it was only present in QEMU 4.0.
Report a validation error for both of the users that will ever hit this
code path.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Now that looking up dnsmasq is handled/mocked we can start
checking whether dnsmasq capabilities were built successfully and
error out if that wasn't the case.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
DISCLAIMER: dnsmasq capabilities are empty as of v8.0.0-rc1~145.
In a real environment the dnsmasq capabilities are constructed
using dnsmasqCapsNewFromBinary(). We also have
dnsmasqCapsNewFromBuffer() to bypass checks that real code is
doing and just get capabilities object. The latter is used from
test suite.
However, with a little bit of mocking we can test the real life
code. All that's needed is to simulate dnsmasq's output for
--version and --help and mock a stat() that's done in
dnsmasqCapsRefreshInternal().
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
While it's true that our virCommand subsystem is happy with
non-absolute paths, the dnsmasq capability code is not. It stores
the path to dnsmasq within and makes it accessible via
dnsmasqCapsGetBinaryPath(). While strictly speaking no caller
necessarily needs canonicalized path, let's find dnsmasq once and
cache the result.
Therefore, when constructing the capabilities structure look up
the binary path. If DNSMASQ already contains an absolute path
then virFindFileInPath() will simply return a copy.
With this code in place, the virFileIsExecutable() check can be
removed from dnsmasqCapsRefreshInternal() because
virFindFileInPath() already made sure the binary is executable.
But introducing virFindFileInPath() means we have to mock it in
test suite because dnsmasqCaps are created in
networkxml2conftest.
Moreover, we don't need to check for dnsmasq in configure.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
We don't query any capabilities of dnsmasq. We are only
interested in dnsmasq's version (obtained via 'dnsmasq
--version'). Therefore, there's no point in running 'dnsmasq
--help'. Its output is not processed even.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
The dnsmasqCaps type has its own cleanup function defined and
ready to use via g_autoptr(). Use automatic cleanup instead of
an explicit one.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
If a test binary is executed with an argument then usage
information is printed out (that no arguments are accepted and
what environment variables affect execution). The string is
printed onto stderr but it is not terminated with a newline
character producing not so nice output.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Now that qemu fixed device unplug when JSON syntax is used with -device
we can re-enable the feature.
Since the old capability string representation is condemned by
suggesting filtering it as a workaround we must introduce a new string.
To achieve this the original capability position is renamed to
X_QEMU_CAPS_DEVICE_JSON_BROKEN_HOTPLUG and a new position with the
original name QEMU_CAPS_DEVICE_JSON is introduced to prevent us having
to change the rest of the code.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Update to commit v6.2.0-874-g1cd2ad11d3
Notable changes are:
- added flag noting that use of JSON syntax for -device was fixed
- 'dbus' backend for graphics and character devices added
- virtio-mem added 'node' property
- 'clusters' added to CPU topology
- 'open-timeout' property for NBD protocol backend
- 'wheel-left' and 'wheel-right' event types for 'input-send-event'
- increased default resolution to '1280x800' on 'virtio-gpu'
- SGX property 'section-size' changed to 'sections' incompatibly
(unused luckily)
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The machine type doesn't change the test result and prevents tests being
changed every time we are about to update real capabilities to a new
qemu.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
There are two tests currently that simulate QMP talk:
qemucapabilitiestest and qemuhotplugtest. In both cases they
check whether currently executed command is the one for which
reply was provided. If not an error message is reported. However,
the error message contains only the actual command and not the
expected one. This makes it harder to navigate through .replies
files.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
VM XML accepts target.port but this does not get passed while
building the QEMU command line for this VM.
Signed-off-by: Divya Garg <divya.garg@nutanix.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This commit takes care of following cases:
-> Check availability of requested ports.
->The total number of requested ports should not be more than
VIR_MAX_ISA_SERIAL_PORTS.
->The ports requested should be less than VIR_MAX_ISA_SERIAL_PORTS.
->VIR_MAX_ISA_SERIAL_PORTS should correspond to MAX_ISA_SERIAL_PORTS
specified in QEMU code commit def337ffda34d331404bd7f1a42726b71500df22.
-> Prevent duplicate device assignments to the same port.
-> In case no ports are provided in the XML, this patch scans the list of unused
isa-serial indices to automatically assign available ports for this VM.
Signed-off-by: Divya Garg <divya.garg@nutanix.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
When -device is configured via JSON a bug [1] is triggered in qemu were
the DEVICE_DELETED event for the removal of the device frontend is no
longer delivered to libvirt. Without the DEVICE_DELETED event we don't
remove the corresponding entries in the VM XML.
Until qemu will be fixed we must stop using the JSON syntax for -device.
This patch removes the detection of the capability. The capability is
used only during startup of a fresh VM so we don't need to consider any
compaitibility steps for existing VMs.
For users who wish to use 'libvirt-7.9' and 'libvirt-7.10' with
'qemu-6.2' there are two possible workarounds:
- filter out the 'device.json' qemu capability '/etc/libvirt/qemu.conf':
capability_filters = [ "device.json" ]
- filter out the 'device.json' qemu capability via qemu namespace XML:
<domain type='kvm' xmlns:qemu='http://libvirt.org/schemas/domain/qemu/1.0'>
[...]
<qemu:capabilities>
<qemu:del capability='device.json'/>
</qemu:capabilities>
</domain>
We must never again use the same capability name as we are now
instructing users to filter it as a workaround so once qemu is fixed
we'll need to pick a new capability value for it.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=2036669
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2035237
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ani Sinha <ani@anisinha.ca>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Until now we did 2 weird things when inserting the qemuCaps used for
individual test cases into the capability cache:
1) we inserted the same caps for all emulators
2) we always (expensively) copied them
Now when real capabilities are used we don't touch them at all just
simply inser them. This allows us one big optimization, by trading a
copy for just a virObjectRef as we can borrow the caps object to the
cache.
For fake caps we still copy them as we insert the fake machine types
into them, but second big optimization is to insert the capabilities
only for the architecture they belong to.
Additionally this commit also ensures that all other entries in the
cache for the binary are poisoned by empty caps so that it's obvious
that the test is doing the right thing.
Apart from this making actually more sense this shaves off more than 40%
of runtime from qemuxml2argvtest.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Emulator binary change is needed to use the latest caps properly. The
comment is no longer needed, the expected error is recorded in the 'err'
file.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
