We're using VIR_AUTOPTR() for everything now, plus the
cleanup section was not doing anything useful anyway.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
In two out of three scenarios we are cleaning up properly after
ourselves, but commit 5f2212c062c7 has changed the remaining one
in a way that caused it to start leaking cpumapToSet.
Refactor the logic so that cpumapToSet is always a freshly
allocated bitmap that gets cleaned up automatically thanks to
VIR_AUTOPTR(); this also allows us to remove the hostcpumap
variable.
Reported-by: John Ferlan <jferlan@redhat.com>
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Ever since the feature was introduced with commit 0f8e7ae33ace,
it has contained a logic error in that it attempted to use a NUMA
node map where a CPU map was expected.
Because of that, guests using <numatune> might fail to start:
# virsh start guest
error: Failed to start domain guest
error: cannot set CPU affinity on process 40055: Invalid argument
This was particularly easy to trigger on POWER 8 machines, where
secondary threads always show up as offline in the host: having
<numatune>
<memory mode='strict' placement='static' nodeset='1'/>
</numatune>
in the guest configuration, for example, would result in libvirt
trying to set the process affinity so that it would prefer
running on CPU 1, but since that's a secondary thread and thus
shows up as offline, the operation would fail, and so would
starting the guest.
Use the newly introduced virNumaNodesetToCPUset() to convert the
NUMA node map to a CPU map, which in the example above would be
48,56,64,72,80,88 - a valid input for virProcessSetAffinity().
https://bugzilla.redhat.com/show_bug.cgi?id=1703661
Signed-off-by: Andrea Bolognani <abologna@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>
If the scheduler is set before vCPU0 cannot be moved into its cpu,cpuacct
cgroup. While it is not yet known whether this is a bug or not, it makes sense
for us to do that later as otherwise the scheduler would be inherited by vCPU
and I/O Threads even when they do not have any such setting specified.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
It's funny how this went unnoticed for such a long time. Long
story short, if a domain is configured with
VIR_DOMAIN_NUMATUNE_MEM_STRICT libvirt doesn't really honour
that. This is because of 7e72ac787848 after which libvirt allowed
qemu to allocate memory just anywhere and only after that it used
some magic involving cpuset.memory_migrate and cpuset.mems to
move the memory to desired NUMA nodes. This was done in order to
work around some KVM bug where KVM would fail if there wasn't a
DMA zone available on the NUMA node. Well, while the work around
might stopped libvirt tickling the KVM bug it also caused a bug
on libvirt side: if there is not enough memory on configured NUMA
node(s) then any attempt to start a domain must fail. Because of
the way we play with guest memory domains can start just happily.
The solution is to move the child we've just forked into emulator
cgroup, set up cpuset.mems and exec() qemu only after that.
This basically reverts 7e72ac787848b7434c9 which was a workaround
for kernel bug. This bug was apparently fixed because I've tested
this successfully with recent kernel.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
The call to resolve the actual network type will turn any NICs with
type=network into one of the other types. Thus there should be no need
to handle type=network in later switch() statements jumping off the
actual type.
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The APIs for allocating/notifying/removing network ports just take
an internal domain interface struct right now. As a step towards
turning these into public facing APIs, add a virNetworkPtr argument
to all of them.
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The port allocation APIs are currently called unconditionally for all
types of NIC, but (mostly) only do anything for NICs with type=network.
The exception is the port allocate API which does some validation even
for NICs with type!=network. Relying on this validation is flawed,
however, since the network driver may not even be installed. IOW virt
drivers must not delegate validation to the network driver for NICs
with type != network.
This change allows us to report errors when the virtual network driver
is not registered.
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
This helps in a scenarios where vCPUs run with a priority that is so high they
might starve the emulator thread. And it also fits with the rest of the
settings.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This does not cause a problem in usual scenarios thanks to us allowing
CAP_DAC_OVERRIDE for the qemu process, however in some scenarios this might be
an issue because the directory is created with mkdtemp(3) which explicitly
creates that with 0700 permissions and qemu running as non-root cannot access
that.
The scenarios include:
- Builds without CAPNG
- Running libvirtd in certain container configurations [1]
- and possibly others.
[1] https://github.com/kubevirt/kubevirt/pull/2181#issuecomment-481840304
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Since the STOP event handler can use the pausedReason as sent to
qemuProcessStopCPUs, we no longer need to send duplicate suspended
lifecycle events because we know what caused the stop along with extra
details. This processing allows us to also remove the duplicated state
change from qemuProcessStopCPUs.
Reviewed-by: John Ferlan <jferlan@redhat.com>
Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>
Map is based on existing cases in code where we send suspended
event after changing domain state to paused.
Reviewed-by: John Ferlan <jferlan@redhat.com>
Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>
Similar to commit [1] which saves and passes the running reason to
the RESUME event handler, during qemuProcessStopCPUs let's save and pass
the pause reason in the domain private data so that the STOP event
handler can use it.
[1] 5dab984ed : qemu: Pass running reason to RESUME event handler
Reviewed-by: John Ferlan <jferlan@redhat.com>
Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>
Now that the core of SnapshotObj is agnostic to snapshots and can be
shared with upcoming checkpoint code, it is time to rename the struct
and the functions specific to list operations. A later patch will
shuffle which file holds the common code. This is a fairly mechanical
patch.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
Replace virDomainChrSourceDefFree with virObjectUnref.
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Use refcounting for priv->monConfig instead of asymmetric freeing.
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Ján Tomko <jtomko@redhat.com>
qemuProcessQMPStart starts a QEMU process and monitor connection that
can be used by multiple functions possibly for multiple QMP commands.
The QMP exchange to exit capabilities negotiation mode and enter command
mode can only be performed once after the monitor connection is
established.
Move responsibility for entering QMP command mode into the
qemuProcessQMP code so multiple functions can issue QMP commands in
arbitrary orders.
This also simplifies the functions using the connection provided by
qemuProcessQMPStart to issue QMP commands.
Test code now needs to call qemuMonitorSetCapabilities to send the
message to switch to command mode because the test code does not use the
qemuProcessQMP command that internally calls qemuMonitorSetCapabilities.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Multiple QEMU processes for QMP commands can operate concurrently.
Use a unique directory under libDir for each QEMU process to avoid
pidfile and unix socket collision between processes.
The pid file name is changed from "capabilities.pidfile" to "qmp.pid"
because we no longer need to avoid a possible clash with a qemu domain
called "capabilities" now that the processes artifacts are stored in
their own unique temporary directories.
"Capabilities" was changed to "qmp" in the pid file name because these
processes are no longer specific to the capabilities usecase and are
more generic in terms of being used for any general purpose QMP message
exchanges with a QEMU process that is not associated with a domain.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Users qemuProcessQMP struct were always forced to call both
qemuProcessQMPStop and qemuProcessQMPFree when they are done with the
process. We can just call qemuProcessQMPStop from qemuProcessQMPFree and
let users call qemuProcessQMPFree only.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
qemuProcessQMPNew is one of the public functions used to create and
manage a QEMU process for QMP command exchanges outside of domain
operations.
Add descriptive comment block, debug statement and make source
consistent with the cleanup / VIR_STEAL_PTR format used elsewhere.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The monitor config data is removed from the qemuProcessQMP struct.
The monitor config data can be initialized immediately before call to
qemuMonitorOpen and does not need to be maintained after the call
because qemuMonitorOpen copies any strings it needs.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Move code for setting paths and prepping file system from
qemuProcessQMPNew to qemuProcessQMPInit.
This keeps qemuProcessQMPNew limited to data structures and path
initialization is done in qemuProcessQMPInit.
The patch is a non-functional, cut / paste change, however goto is now
"cleanup" rather than "error".
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Store libDir path in the qemuProcessQMP struct in anticipation of moving
path construction code into qemuProcessQMPInit function.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
All code related to QEMU monitor is moved from qemuProcessQMPNew and
qemuProcessQMPInit into qemuProcessQMPConnectMonitor.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This is a replacement for qemuProcessQMPRun to make the name consistent
with qemuProcessStart. The original qemuProcessQMPRun function is
renamed as qemuProcessQMPLaunch and becomes one of the simpler functions
called from the main qemuProcessQMPStart entry point. The following
patches will move parts of the code in qemuProcessQMPLaunch to the other
functions (qemuProcessQMPInit and qemuProcessQMPConnectMonitor).
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Keep the pointer to QEMU stderr output in qemuProcessQMP struct instead
of requiring the caller to provide it (and free it).
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
While qemuProcessQMPRun and virQEMUCapsInitQMPMonitor* functions called
from virQEMUCapsInit ignore some errors, the caller of virQEMUCapsInit
would report an error unless usedQMP is true anyway. And since usedQMP
can only be true if the probing code really succeeded (i.e., no errors
were ignored), we can just simplify the logic by not ignoring the errors
in the first place.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
In new process code, move from model where qemuProcessQMP struct can be
used to activate a series of Qemu processes to model where one
qemuProcessQMP struct is used for one and only one Qemu process.
By allowing only one process activation per qemuProcessQMP struct, the
struct can safely store process outputs like status and stderr, without
being overwritten, until qemuProcessQMPFree is called.
By doing this, process outputs like status and stderr can remain stored
in the qemuProcessQMP struct without being overwritten by subsequent
process activations.
The forceTCG parameter (use / don't use KVM) will be passed when the
qemuProcessQMP struct is initialized since the qemuProcessQMP struct
won't be reused.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
virQEMUCapsInitQMP now stops QEMU process in all execution paths,
before freeing the process structure.
The qemuProcessQMPStop function can be called multiple times without
problems... Won't attempt to stop processes and free resources multiple
times.
Follow the convention established in qemu_process of
1) alloc process structure
2) start process
3) use process
4) stop process
5) free process data structure
The process data structure persists after the process activation fails
or the process dies or is killed so stderr strings can be retrieved
until the process data structure is freed.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
s/qemuProcessQMPAbort/qemuProcessQMPStop/ applied to change function
name used to stop QEMU processes in process code moved from
qemu_capabilities.
No functionality change.
The new name, qemuProcessQMPStop, is consistent with the existing
function qemuProcessStop used to stop Domain processes.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
s/cmd/proc/ in process code imported from qemu_capabilities.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Add the const qualifier on non modified strings
(string only copied inside qemuProcessQMPNew)
so that const strings can be used directly in calls to
qemuProcessQMPNew in future patches.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
QEMU process code in qemu_capabilities.c is moved to qemu_process.c in
order to make the code usable outside the original capabilities use
cases.
The moved code activates and manages QEMU processes without establishing
a guest domain.
This patch is a straight cut/paste move between files.
Signed-off-by: Chris Venteicher <cventeic@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
For normal starts (no incoming migration) the refresh of the QEMU
state must be done before the VCPUs getting started since otherwise
there might be a race condition between a possible shutdown of the
guest OS and the QEMU monitor queries.
This fixes "qemu: migration: Refresh device information after
transferring state" (93db7eea1b864).
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1503284
The way we currently start qemu from CPU affinity POV is as
follows:
1) the child process is set affinity to all online CPUs (unless
some vcpu pinning was given in the domain XML)
2) Once qemu is running, cpuset cgroup is configured taking
memory pinning into account
Problem is that we let qemu allocate its memory just anywhere in
1) and then rely in 2) to be able to move the memory to
configured NUMA nodes. This might not be always possible (e.g.
qemu might lock some parts of its memory) and is very suboptimal
(copying large memory between NUMA nodes takes significant amount
of time).
The solution is to set affinity to one of (in priority order):
- The CPUs associated with NUMA memory affinity mask
- The CPUs associated with emulator pinning
- All online host CPUs
Later (once QEMU has allocated its memory) we then change this
again to (again in priority order):
- The CPUs associated with emulator pinning
- The CPUs returned by numad
- The CPUs associated with vCPU pinning
- All online host CPUs
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Be more sensible when setting labels of the target of a
virDomainBlockCopy operation. Previously we'd relabel everything in case
it's a copy job even if there's no unlabelled backing chain. Since we
are also not sure whether the backing chain is shared we don't relabel
the chain on completion of the blockjob. This certainly won't play nice
with the image permission relabelling feature.
While this does not fix the case where the image is reused and has
backing chain it certainly sanitizes all the other cases. Later on it
will also allow to do the correct thing in cases where only one layer
was introduced.
The change is necessary as in case when -blockdev will be used we will
need to hotplug the backing chain and thus labeling needs to be setup in
advance and not only at the time of pivot. To avoid multiple code paths
move the labeling now.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
When we need to detect a chain for a image which will become the new
source for a disk (e.g. after a disk media change or a blockjob) we'd
need to replace disk->src temporarily to do so.
Move the 'disksrc' temporary variable to an argument and adjust callers.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
If we validate that memballoon is NONE|VIRTIO at parse time,
we can drop similar checks elsewhere in the qemu driver
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Hanlde all the possible failure codes as per ACPI standard documented in
the function header.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1660410
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
We forgot to document the specific fields for the 0x103 and 0x200
sources which are tied to device removal and device hotplug
respectively.
The value description is based on the ACPI 6.2A standard Table 6-207 and
Table 6-208. At the time of writing of this patch the standard can be
accessed e.g. at:
https://www.uefi.org/sites/default/files/resources/ACPI%206_2_A_Sept29.pdf
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Be generic instead of trying to enumerate all the involved
device types.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
Currently the job name corresponds to the disk the job belongs to. For
jobs which will not correspond to disks we'll need to track the name
separately.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Add a field tracking the current state of job so that it can be queried
later. Until now the job state e.g. that the job is _READY for
finalizing was tracked only for mirror jobs. Add tracking of state for
all jobs.
Similarly to 'qemuBlockJobType' this maps the existing states of the
blockjob from virConnectDomainEventBlockJobStatus to
'qemuBlockJobState' so that we can track some internal states as well.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
