Commit 24be92b8e moved the option rom settings validation code to the
validation callbacks, but that doesn't work properly with device hotplug
as we assign addresses only after parsing the whole XML. The check is
too strict for that and caused failures when hotplugging devices such
as:
<interface type='network'>
<source network='default'/>
<model type='virtio'/>
<rom enabled='no'/>
</interface>
This patch relaxes the check in the validation callback to accept also
_NONE and _UNASSIGNED address types and returns the check to
'qemuBuildRomProps' so that we preserve the full validation as we've
used to.
Fixes: 24be92b8e38762e9ba13e
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2021437
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We currently use -machine accel=XXX which is just a syntax sugar
for -accel XXX. The former doesn't allow specifying arguments for
accelerator, because all arguments passed to -machine are
treated as arguments of machine itself.
The -accel argument was introduced in QEMU commit
v2.9.0-rc0~70^2~19 and since our minimum required version is
newer (2.11.0) we can safely assume its existence and use it
without any capability.
Resolves: https://gitlab.com/libvirt/libvirt/-/issues/233
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Tested-by: Kashyap Chamarthy <kchamart@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
The domain accelerator was validated in qemuValidateDomainDef()
which calls virQEMUCapsIsVirtTypeSupported() which reports proper
error if QEMU is not capable of KVM/TCG. There is no point in
doing the validation again when building command line.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Tested-by: Kashyap Chamarthy <kchamart@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
When guest has NUMA nodes and QEMU is new enough to report
default RAM ID then ideally we would use -numa memdev= combined
with memory-backend-* combo becasue -mem-path/-mem-prealloc/-numa
mem are deprecated. Well, there is one problem - the .memdev=
attribute is machine type dependent (just look at arguments of
virQEMUCapsGetMachineNumaMemSupported()) and to ensure backwards
compatibility we prefer -numa mem= over -numa memdev=.
But there was one corner case when -mem-prealloc was requested
but not generated on the cmd line. It all starts with
qemuBuildMemCommandLine() which generates just '-m XXX' and
because it sees defaultRAMid and guest NUMA nodes greater than
zero it does nothing more.
Then, qemuBuildNumaCommandLine() sees that -numa mem= is still
supported for given machine type and nothing else set
@needBackend thus qemuBuildMemPathStr() is called which output
-mem-prealloc only in a few cases assuming it was outputted
earlier.
Reported-by: Jing Qi <jinqi@redhat.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Before commit 73c352ab8c97d3 which converted the hostdev commandline
formatter to JSON the 'ramfb' property was formatted only if it was
enabled.
The main reason for that is that enabling 'ramfb' switches the device
model to 'vfio-pci-nohotplug' which actually has the property, while
'vfio-pci' (used when 'ramfb' is disabled or absent) doesn't have it.
Restore the logic to format 'ramfb' only when it's enabled and add a
comment that it's deliberate.
Fixes: 73c352ab8c97d3
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2024435
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'virDomainChrSourceDef' contains private data so 'virDomainChrSourceDefNew'
must be used to allocate it. 'virDomainTPMDef' was using it directly
which won't work with the chardev helper functions.
Convert it to a pointer to properly allocate private data.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The main reason is to ensure that the private data are properly
allocated for every instance.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The function is already returning JSON properties, rename it
accordingly.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'-audiodev' as a modern implementation based on QAPI already takes JSON
as the argument. Convert our code to use it directly.
The declaration of the QAPI types can be found in
'qemu.git/qapi/audio.json'.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
virJSONValueObjectAdd now works identically to virJSONValueObjectCreate
when used with a NULL argument. Replace all callers.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Upcomming patch will convert virJSONValueObjectCreate which didn't check
the value of the passed argument to virJSONValueObjectAdd which will and
this would lead to crashes.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Change the first argument to a double pointer so that later the function
can be unified with virJSONValueObjectCreate and fix all callers.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Commit 88957116c9d3cb4705380c3702c9d4315fb500bb switched to use
memory-backend-* for regular VM memory as well. That change indirectly
started using 'host-nodes' for system memory which results in QEMU
calling mbind() to bind the system memory to specific NUMA node if the
VM XML contains the configuration similar to this:
...
<numatune>
<memory mode='strict' nodeset='0'/>
</numatune>
...
Once the VM was started with that configuration it was no longer
possible to change the memory NUMA nodeset.
Fixes: 677c90cc1d1fcb3aba09b5d4f0f8f83099911775
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The return value was not properly checked in the code for input type
evdev.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
QEMU version 3.1 introduced PV_SEND_IPI CPUID feature bit under
commit 7f710c32bb8 (target-i386: adds PV_SEND_IPI CPUID feature bit).
This patch adds a new KVM feature 'pv-ipi' to disable this feature
(enabled by default). Newer CPU platform (Ex, AMD Zen2) supports
hardware accelation for IPI in guest, to use this feature to get
better performance in some scenarios. Detailed about the discussion:
https://lkml.org/lkml/2021/10/20/423
To disable kvm-pv-ipi and have libvirt add "-cpu host,kvm-pv-ipi=off"
to the QEMU command line, the following XML code needs to be added to the
guest's domain description:
<features>
<kvm>
<pv-ipi state='off'/>
</kvm>
</features>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
qemuBuildPMPCIRootHotplugCommandLine() returns 0 unconditionally. There is no
failure scenario at present. So clean up the code by removing integer return
from the function and also remove the failure check conditional from the
function call.
Also fix indentation for the above function call while at it.
Signed-off-by: Ani Sinha <ani@anisinha.ca>
Reviewed-by: Laine Stump <laine@redhat.com>
This reverts commit bef0f0d8be6baa1d9359be208b53d6b8a37ddc95.
Conflicts:
tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-disable.args
* this file had been renamed from its original, then renamed back,
which understandably confused git. It's being completely removed
here anyway, so the contents don't matter.
tests/qemuxml2argvtest.c
* change in context around removed chunk
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Ani Sinha <ani@anisinha.ca>
This reverts commit 618e8665db2e4c1a8e9a227045b99b48f6110c06.
This is the first in a series of 10 commits that revert (in reverse
order) the changes to add the <acpi-hotplug-bridge state='on|off'/>
switch to libvirt domain XML, which unfortunately needs to be removed
due to QEMU developers discovering a flaw with the design of the QEMU
commandline switch used to implement the libvirt switch that will
likely result in a new and different method of selecting hotplug
modes. Because the libvirt switch has not been in any official
releases of libvirt, we are still able to remove it completely, rather
than deprecating it.
The original commits began with commit
58ba0f6a3d7342fba29edbbf2bb9cb5497c870e5. The other original commit
IDs are documented in each revert commit.
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Ani Sinha <ani@anisinha.ca>
In a few places we declare a variable (which is optionally
followed by a code not touching it) then set the variable to a
value and return the variable immediately. It's obvious that the
variable is needless and the value can be returned directly
instead.
This patch was generated using this semantic patch:
@@
type T;
identifier ret;
expression E;
@@
- T ret;
... when != ret
when strict
- ret = E;
- return ret;
+ return E;
After that I fixed couple of formatting issues because coccinelle
formatted some lines differently than our coding style.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This involves a bit of a hack, but is overall preferable to
forcing callers to pass non-const devdata as argument.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Some guest features that map to the -cpu arg are still added using
implicit syntax "feature" which is a deprecated shorthand for
"feature=on".
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
With the previous refactorings, there's no real benefit from the
qemuBuildCpuFeature helper method. Only one of the callers really
needs the CPU feature name re-writing logic, the others can just
use the right name directly.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The -cpu arg gained support for feature=on|off syntax for the x86
emulator in 2.4.0
commit 38e5c119c2925812bd441450ab9e5e00fc79e662
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Mon Mar 23 17:29:32 2015 -0300
target-i386: Register QOM properties for feature flags
Most other targets gained this syntax even earlier in 1.4.1
commit 1590bbcb02921dfe8e3cf66e3a3aafd31193babf
Author: Andreas Färber <afaerber@suse.de>
Date: Mon Mar 3 23:33:51 2014 +0100
cpu: Implement CPUClass::parse_features() for the rest of CPUs
CPUs who do not provide their own implementation of feature parsing
will treat each option as a QOM property and set it to the supplied
value.
There appears no reason to keep supporting "+|-feature" syntax,
given the current minimum QEMU version.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
QEMU switched from using underscores in x86 CPU features to hyphens
in the 2.8.0 series with two commits
commit fc7dfd205f3287893c436d932a167bffa30579c8 (HEAD, refs/bisect/bad)
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Fri Sep 30 15:49:40 2016 -0300
target-i386: Remove underscores from feat_names arrays
commit 54b8dc7c19cd781e96f1e9b001ca6001d804eb19
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Fri Sep 30 15:49:38 2016 -0300
target-i386: Register aliases for feature names with underscores
Libvirt names use underscores so we conditionally tranlate the
names when talking to new QEMU. Since the min QEMU was raised to
version 2.11.0, all QEMU versions we talk to expect hypens, so
the translation can be done unconditionally.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
QEMU declares the bootindex types as:
bootindexA=<int32>
bootindexB=<int32>
The driveA/driveB parameters were deprecated and removed in qemu-6.0.
We'll keep them for compatibility, but they are not used with -blockdev.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
qemuBuildFloppyCommandLineControllerOptions was generating config for
both the implicit and explicit fdc. The explicit FDC is using '-device'
and thus will need to be converted to JSON.
Split up the lookup of the floppy drive configs from the actual command
generation.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
QEMU declares the props we control as:
'ccid-card-emulated'
backend=<str>
cert1=<str>
cert2=<str>
cert3=<str>
db=<str>
'ccid-card-passthru'
chardev=<str> - ID of a chardev to use as a backend
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
While this device doesn't have any properties it must be converted to
use qemuBuildDeviceCommandlineFromJSON so that we can validate it in the
future.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
QEMU declares the 'guid' property as:
guid=<str> - UUID (aka GUID) or "auto" for random value (default) (default: "auto")
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Move it into the validator. Note that the placement into the device
validation part is intentional so that it also covers hotplug code
paths.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We have a commonly used helper virDomainControllerAliasFind, which does
the same thing and also reports errors internally.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Use 'virXMLPropEnum' to parse it and fix all switch statements which
didn't include the VIR_DOMAIN_SMARTCARD_TYPE_LAST case.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Now that the code was converted to use this helper we can remove it.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We control only the 'tpmdev' property of TPM devices which is a string.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The codec devices have the following properties we control:
cad=<uint32> - (default: 4294967295)
audiodev=<str> - ID of an audiodev to use as a backend
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The sound devices have only the 'audiodev' property which is a string.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
All virtio devices were converted to the new JSON formatter so we can
remove the old one.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We control the following properties of the devices in question:
'virtio-gpu'
virgl=<bool> - on/off (default: true)
'qxl'
ram_size=<uint32> - (default: 67108864)
vram_size=<uint64> - (default: 67108864)
vram64_size_mb=<uint32> - (default: 4294967295)
vgamem_mb=<uint32> - (default: 16)
max_outputs=<uint16> - (default: 0)
'vhost-user-gpu'
max_outputs=<uint32> - (default: 1)
chardev=<string>
'VGA'
vgamem_mb=<uint32> - (default: 16)
'bochs-display'
vgamem=<size> - (default: 16777216)
common for all devices:
xres=<uint32> - (default: 0)
yres=<uint32> - (default: 0)
The only noticable change is using memory size in bytes for
'bochs-display' instead of kibibytes.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
