The pconfig feature was enabled in QEMU by accident in 3.1.0. All other
newer versions do not support it and it was removed from the
Icelake-Server CPU model in QEMU.
We don't normally change our CPU models even when QEMU does so to avoid
breaking migrations between different versions of libvirt. But we can
safely do so in this specific case. QEMU never supported enabling
pconfig so any domain which was able to start has pconfig disabled.
With a small compatibility hack which explicitly disables pconfig when
CPU model equals Icelake-Server in migratable domain definition, only
one migration scenario stays broken (and there's nothing we can do about
it): from any host to a host with libvirt < 5.10.0 and QEMU > 3.1.0.
https://bugzilla.redhat.com/show_bug.cgi?id=1749672
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
QEMU does not support setting this feature on the command line anymore.
We don't need to explain why it is not included in CPU models then.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
When a CPU definition wants to explicitly disable some features that are
unknown to QEMU, we can safely drop them from the definition before
starting QEMU. Naturally QEMU won't enable such features implicitly.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
QEMU 2.11 for ppc64 changed all CPU model names to lower case. Since
libvirt can't change the model names for compatibility reasons, we need
to translate the matching lower case models to the names known by
libvirt.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This reverts commit 2d8721e260.
This fix was both incomplete and too general. It only fixed domain
startup, but libvirt would still report empty list of supported CPU
models with recent QEMU for ppc64. On the other hand, while ppc64 QEMU
ignores case when looking up CPU model names, x86_64 QEMU does case
sensitive lookup. Without reverting this patch, libvirt could happily
accept CPU model names which are not supported by QEMU.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Once we copy the domain definition from virDomainSnapshotDef, we either
need to assign it to the domain object or free it to avoid memory leaks.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
The only caller for which this check makes sense is virDomainDefParse.
Thus the check should be moved there.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@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: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The (pre-copy) bandwidth was historically the only bandwidth we
supported and thus it is called just "bandwidth" in all other places.
E.g., virsh migrate-setspeed or in the migration typed parameter name.
Let's make the new option for virsh migrate consistent.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
The (pre-copy) bandwidth was historically the only bandwidth we
supported and thus it is called just "bandwidth" in all other places.
E.g., virsh migrate-setspeed or in the migration typed parameter name.
Let's make the new option for virsh migrate consistent.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
When QEMU supports flushing caches at the end of migration, we can
safely allow migration even if disk/driver/@cache is not none nor
directsync.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Acked-By: Peter Krempa <pkrempa@redhat.com>
QEMU 4.0.0 and newer automatically drops caches at the end of migration.
Let's check for this capability so that we can allow migration when disk
cache is turned on.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Acked-By: Peter Krempa <pkrempa@redhat.com>
The original message was logically incorrect: cache != none or cache !=
directsync is always true. But even replacing "or" with "and" doesn't
make it more readable for humans.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Acked-By: Peter Krempa <pkrempa@redhat.com>
In the first stage of incoming migration (qemuMigrationDstPrepareAny) we
call qemuMigrationEatCookie when there's no vm object created yet and
thus we don't have any private data to pass.
Broken by me in commit v5.6.0-109-gbf15b145ec.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
This reverts commit 47cbc92987.
The section is no longer correct when the patch switching to gnulib's
make coverage was reverted.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Acked-By: Eric Blake <eblake@redhat.com>
This reverts commit f38d553e2d.
Gnulib's make coverage (or init-coverage, build-coverage, gen-coverage)
is not a 1-1 replacement for the original configure option. Our old
--enable-test-coverage seems to be close to gnulib's make build-coverage
except gnulib runs lcov in that phase and the build actually fails for
me even before lcov is run. And since we want to be able to just build
libvirt without running lcov, I suggest reverting to our own
implementation.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Acked-By: Eric Blake <eblake@redhat.com>
Since qemuDomainDeviceDefPostParse callback requires qemuCaps, we need
to make sure it gets the capabilities stored in the domain's private
data if the domain is running. Passing NULL may cause QEMU capabilities
probing to be triggered in case QEMU binary changed in the meantime.
When this happens while a running domain object is locked, QMP event
delivered to the domain before QEMU capabilities probing finishes will
deadlock the event loop.
QEMU capabilities lookup (via domainPostParseDataAlloc callback) is
hidden inside virDomainDeviceDefPostParseOne with no way to pass
qemuCaps to virDomainDeviceDef* functions. This patch fixes all
remaining paths leading to virDomainDeviceDefPostParse.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
Several general snapshot and checkpoint APIs were lazily passing NULL as
the parseOpaque pointer instead of letting their callers pass the right
data. This patch fixes all paths leading to virDomainDefParseNode.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to virDomainDefPostParse.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
Several general functions from domain_conf.c were lazily passing NULL as
the parseOpaque pointer instead of letting their callers pass the right
data. This patch fixes all paths leading to virDomainDefCopy to do the
right thing.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to qemuMigrationCookieXMLParse.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to virDomainDefParseString.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to qemuMigrationAnyPrepareDef.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to qemuDomainSaveImageOpen.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to qemuDomainDefFormatBufInternal.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Since qemuDomainDefPostParse callback requires qemuCaps, we need to make
sure it gets the capabilities stored in the domain's private data if the
domain is running. Passing NULL may cause QEMU capabilities probing to
be triggered in case QEMU binary changed in the meantime. When this
happens while a running domain object is locked, QMP event delivered to
the domain before QEMU capabilities probing finishes will deadlock the
event loop.
This patch fixes all paths leading to qemuDomainDefCopy.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
hv-spinlocks is not a CPUID feature and should not be checked as such.
While starting a domain with hv-spinlocks enabled, we would report a
warning about unsupported hyperv spinlocks feature even though it was
set properly.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Originally the names of the KVM CPU features were only used internally
for looking up their CPUID bits. So we used "__kvm_" prefix for them to
make sure the names do not collide with normal CPU features stored in
our CPU map.
But with QEMU 4.1 we check which features were enabled or disabled by a
freshly started QEMU process using their names rather than their CPUID
bits (mostly because of MSR features). Thus we need to change our made
up internal names into the actual names used by QEMU.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Tested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Most of the internally defined KVM CPUID features are not actually used
by libvirt. The QEMU driver may enable or disable them on the command
line, but we don't check for the associated CPU properties or CPUID
bits. They would be useless with QEMU 4.1 anyway since their names were
only remotely similar to the actual feature names.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Tested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
All the features are hyperv features even though they are provided by
KVM with QEMU. The "KVM" part in the macro names does not make a lot of
sense.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Tested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Starting with QEMU 4.1, we're using the canonical feature names on the
command line and avoid aliases to prepare for possible deprecation of
all aliases in QEMU. But we do so only for features from our CPU map,
hyperv features defined in the code were unchanged and this patch fixes
it. Some features use "hv-" prefix unconditionally because they were
introduced recently enough to always support spelling with a dash.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Tested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Originally the names of the hyperv CPU features were only used
internally for looking up their CPUID bits. So we used "__kvm_hv_"
prefix for them to make sure the names do not collide with normal CPU
features stored in our CPU map.
But with QEMU 4.1 we check which features were enabled or disabled by a
freshly started QEMU process using their names rather than their CPUID
bits (mostly because of MSR features). Thus we need to change our made
up internal names into the actual names used by QEMU. Most of the names
are only used with QEMU 4.1 and newer and the reset was introduced with
QEMU recently enough to already support spelling with "-". Thus we don't
need to define them as "hv_*" with a translation to "hv-*" for new QEMU.
Without this patch libvirt would mistakenly report all hyperv features
as unavailable and refuse to start any domain using them with QEMU 4.1.
Reported-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Tested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Starting with QEMU 4.1 qemuMonitorCPUModelInfo structure in virQEMUCaps
stores only canonical feature names which may differ from the name used
by libvirt. We need translate these canonical names into libvirt names
for further consumption.
This fixes a bug in qemuConnectBaselineHypervisorCPU which would remove
all features for which libvirt's spelling differs from the QEMU's
preferred name. For example, the following result of
qemuConnectBaselineHypervisorCPU on my host with QEMU 4.1 is wrong:
<cpu mode='custom' match='exact'>
<model fallback='forbid'>Skylake-Client</model>
<vendor>Intel</vendor>
<feature policy='require' name='ss'/>
<feature policy='require' name='vmx'/>
<feature policy='require' name='hypervisor'/>
<feature policy='require' name='clflushopt'/>
<feature policy='require' name='umip'/>
<feature policy='require' name='arch-capabilities'/>
<feature policy='require' name='xsaves'/>
<feature policy='require' name='pdpe1gb'/>
<feature policy='require' name='invtsc'/>
<feature policy='disable' name='pclmuldq'/>
<feature policy='disable' name='lahf_lm'/>
</cpu>
The 'pclmuldq' and 'lahf_lm' should not be disabled in the baseline CPU
as they are supported by QEMU on this host.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Ever since --parallel-connections option for virsh migrate was
introduced we did not properly check the return value of
vshCommandOptInt. We would set VIR_MIGRATE_PARAM_PARALLEL_CONNECTIONS
parameter even if vshCommandOptInt returned 0 (which means
--parallel-connections was not specified) when another int option which
was checked earlier was specified with a nonzero value.
Specifically, running virsh migrate with either
--auto-converge-increment, --auto-converge-initial, --comp-mt-dthreads,
--comp-mt-threads, or --comp-mt-level would set
VIR_MIGRATE_PARAM_PARALLEL_CONNECTIONS parameter and if --parallel
option was not used, libvirt would complain
error: invalid argument: Turn parallel migration on to tune it
even though --parallel-connections option was not used at all.
https://bugzilla.redhat.com/show_bug.cgi?id=1726643
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
With QEMU versions which lack "unavailable-features" we use CPUID based
detection of features which were enabled or disabled once QEMU starts.
Thus using MSR features with host-model would result in all of them
being marked as disabled in the active domain definition even though
QEMU did not actually disable them.
Let's make sure we add MSR features to host-model only when
"unavailable-features" property is supported by QEMU.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Without "unavailable-features" CPU property we cannot properly detect
whether a specific MSR feature we asked for (either explicitly or
implicitly via a CPU model) was disabled by QEMU for some reason.
Because this could break migration, snapshots, and save/restore
operaions, it's better to just forbid any use of MSR features with QEMU
which lacks "unavailable-features" CPU property.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This is used by the host capabilities code to construct host CPU
definition.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This functions may be used as a virCPUDefFeatureFilter callbacks for
virCPUDefCheckFeatures, virCPUDefFilerFeatures, and similar functions to
select (virCPUx86FeatureFilterSelectMSR) or drop
(virCPUx86FeatureFilterDropMSR) features reported via MSR.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Until now, this was a macro usable for direct initialization when a
variable is defined. Turning the macro into a function makes it more
general.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This API can be used to check whether a CPU definition contains features
matching a given filter.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
When updating guest CPU definition according to the vCPU actually
created by QEMU, we want to use the generic qemuMonitorGetGuestCPU to
get both CPUID and MSR features.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Unlike the old version (which is now called qemuMonitorGetGuestCPUx86),
this monitor API checks for individual features by their names rather
than processing CPUID bits. Thus we can get the list of enabled and
disabled features for both CPUID and MSR features.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The function converts a list of QOM properties into a NULL-terminated
array of property names. The new type parameter may be used to limit the
result to properties of a specific type.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This is a generic replacement for the former virCPUx86DataAddFeature,
which worked on the generic virCPUDataPtr anyway.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
It was never implemented or used for anything else anyway. Mainly
because it uses CPUID features bits. The function is renamed as
qemuMonitorGetGuestCPUx86 to make this explicit.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We used type=full expansion on the result of previous type=static
expansion to get all possible spellings of CPU features. Since we can
now translate the QEMU's canonical names to our names, we can drop this
magic and do only type=static CPU model expansion.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
By default query-cpu-model-expansion only reports canonical names of all
CPU features. We do some magic and call the command twice to get all
possible spellings of the features, but being able to consume canonical
names will allow us to drop this magic.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
When building QEMU command line, we should use the preferred spelling of
each CPU feature without relying on compatibility aliases (which may be
removed at some point).
The "unavailable-features" CPU property is used as a witness for the
correct names of the features in our translation table.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The way we call query-cpu-model-expansion will rely on some capabilities
bits. Let's make sure all capabilities are set before probing host CPU.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
It is similar to "filtered-features" property, which reports CPUID bits
corresponding to disabled features, but more general. The
"unavailable-features" property supports both CPUID and MSR features by
listing their names.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We will use it to check whether QEMU supports a specific CPU property.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
So far we always used libvirt's name of each CPU feature relying on
backward compatible aliases in QEMU. The new translation table can be
used whenever QEMU mandates or prefers canonical feature names.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This should cover all CPU features for which QEMU prefers spelling that
differs from the one used by libvirt.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Newer QEMU will translate the feature names to their canonical names so
4.0.0 is the last one which produces the results we currently have in
*-latest.args.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Normal CPU features use modern -cpu ...,feature=on|off syntax when
available, but kvm features kept using the old +feature or -feature.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
These test check all kvm CPU features that could be passed to the -cpu
option by libvirt.
The 2.7.0 version is the last one for which we use +|-feature syntax for
CPU features, while feature=on|off is used with newer versions. This
is visible in the following patch which changes only the *-latest.args
files.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Properly filter features which should not be passed to QEMU because they
were never supported by QEMU or they did nothing and QEMU dropped them.
Currently they are just silently ignored by the command line generator.
Let's make this process more visible and clean by dropping the features
from the domain's active definition in qemuProcessUpdateGuestCPU.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This new internal API can be used for in place filtering of CPU features
in virCPUDef.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We already have virQEMUCapsCPUFilterFeatures for filtering features
which QEMU does not know about. Let's move osxsave and ospke from
qemuFeatureNoEffect there.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Due to this bug the following command would fail on any host where TSC
frequency can be probed:
$ virsh capabilities | virsh cpu-baseline /dev/stdin
error: unsupported configuration: Invalid TSC frequency
https://bugzilla.redhat.com/show_bug.cgi?id=1641702
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
When migrating a domain with invtsc CPU feature enabled, the TSC
frequency of the destination host must match the frequency used when the
domain was started on the source host or the destination host has to
support TSC scaling.
If the frequencies do not match and the destination host does not
support TSC scaling, QEMU will fail to set the right TSC frequency when
starting vCPUs on the destination and thus migration will fail. However,
this is quite late since both host might have spent significant time
transferring memory and perhaps even storage data.
By adding the check to libvirt we can let migration fail before any data
starts to be sent over. If for some reason libvirt is unable to detect
the host's TSC frequency or scaling support, we'll just let QEMU try and
the migration will either succeed or fail later.
Luckily, we mandate TSC frequency to be explicitly set in the domain XML
to even allow migration of domains with invtsc. We can just check
whether the requested frequency is compatible with the current host
before starting QEMU.
https://bugzilla.redhat.com/show_bug.cgi?id=1641702
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
When the host CPU supports invariant TSC the host CPU definition created
by virCPUx86GetHost will contain (unless probing fails for some reason)
addition TSC related data.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Commit 0a97486e09 moved them outside #ifdef, but after virCPUx86GetHost,
which will start calling them in the following patch.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
This patch adds a new
<counter name='tsc' frequency='N' scaling='on|off'/>
element into the host CPU capabilities XML.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
On a KVM x86_64 host which supports invariant TSC this function can be
used to detect the TSC frequency and the availability of TSC scaling.
The magic MSR numbers required to check if VMX scaling is supported on
the host are documented in Volume 3 of the Intel® 64 and IA-32
Architectures Software Developer’s Manual.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
The qemuTestParseCapabilitiesArch call would eventually lead to the host
CPU being probed via virCPUGetHost. Let's divert this to a mocked
version already used by the qemuxml2argvtest.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
The function is renamed as virQEMUCapsProbeHostCPU and it does not get
the list of allowed CPU models from qemuCaps anymore. This is
responsibility is moved to the caller. The result is just a very thin
wrapper around virCPUGetHost mostly required mocking in tests.
The generic function is used in place of a direct call to virCPUGetHost
in virQEMUCapsInitHostCPUModel to make sure tests don't accidentally
probe host CPU.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Depending on the way ctags was compiled, it may look for
.ctags.d/*.ctags files rather than .ctags for reading configuration.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Acked-by: Michal Privoznik <mprivozn@redhat.com>
While libvirtd creates this directory with the default 0755 mode, the
spec file stores 0700 in the RPM database. Thus RPM verification always
complains about this directory. Let's fix the spec file to match
reality.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
The nwfilter XML files stored in /etc/libvirt/nwfilter are copied in a
%post scriptlet from /usr/share/libvirt/nwfilter/*.xml. While the files
in /usr/share are created with mode 0644, libvirt creates the files in
/etc/libvirt/nwfilter with mode 0600. Since 0600 is also stored in the
RPM database, we need to chmod the files copied from /usr/share to make
sure RPM verification does not complain about changed permissions.
https://bugzilla.redhat.com/show_bug.cgi?id=1628475
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
The libvirt-lock-sanlock subpackage requires sanlock to be installed
first and the sanlock package creates the sanlock group on all distros
we care about in the spec file (Fedora and RHEL >= 7). Thus instead of
setting the ownership and permissions in a post scriptlet only when the
sanlock group exists we can just install the directory with the
appropriate metadata.
https://bugzilla.redhat.com/show_bug.cgi?id=1702758
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Acked-by: Michal Privoznik <mprivozn@redhat.com>
CVE-2018-12126, CVE-2018-12127, CVE-2018-12130, CVE-2019-11091
The bit is set when microcode provides the mechanism to invoke a flush
of various exploitable CPU buffers by invoking the VERW instruction.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Added in QEMU by v2.12.0-481-g0da0fb0628 (released in 3.0).
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
My earlier commit be46f61326 was incomplete. It removed caching of
microcode version in the CPU driver, which means the capabilities XML
will see the correct microcode version. But it is also cached in the
QEMU capabilities cache where it is used to detect whether we need to
reprobe QEMU. By missing the second place, the original commit
be46f61326 made the situation even worse since libvirt would report
correct microcode version while still using the old host CPU model
(visible in domain capabilities XML).
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The new virHostCPUGetMSR internal API will try to read the MSR from
/dev/cpu/0/msr and if it is not possible (the device does not exist or
libvirt is running unprivileged), it will fallback to asking KVM for the
MSR using KVM_GET_MSRS ioctl.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This patch adds an inline python code for reading MSR features. Since
reading MSRs is a privileged operation, we have to read them from
/dev/cpu/*/msr if it is readable (i.e., the script runs as root) or
fallback to using KVM ioctl which can be done by any user that can start
virtual machines.
The python code is inlined rather than provided in a separate script
because whenever there's an issue with proper detection of CPU features,
we ask the reporter to run cpu-gather.sh script to give us all data we
need to know about the host CPU. Asking them to run several scripts
would likely result in one of them being ignored or forgotten.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The parseMapFeature for parsing features from CPU map XML can be easily
generalized to support more feature types.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Let's make sure the current CPUID specific code is only applied to CPUID
features.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This will let us simplify the code since the dictionary keys will match
attribute names in various XMLs.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
leaf["eax"] & eax > 0 check works correctly only if there's at most 1
bit set in eax. Luckily that's been always the case, but fixing this
could save us from future surprises.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The function will have to deal with both CPUID and MSR features.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We don't really need to parse CPU data from QEMU older than 2.9 (i.e.,
before query-cpu-model-expansion) at this point. But even if there's a
need to do so, we can always use an older version of this script to do
the conversion.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
They are static and we will need to call them a little bit closer to the
beginning of the file.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The structure can only be used for CPUID data now. Adding a type
indicator and moving the data into a union will let us store alternative
data types.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Jiri Denemark