Commit 01ca4010 (libvirt v5.1.0) moved address reservation for
hotplugged interface devices up to an earlier point in
qemuDomainAttachNetDevice(), because that function calls
qemuDomainSupportsNicdev() (in the case of
VIR_DOMAIN_NET_TYPE_VHOSTUSER), and qemuDomainSupportsNicdev() needs
to know the address type (for ARM machinetypes) and returns incorrect
results when the address type is "none".

This bugfix unfortunately caused a regression, because it also made PCI
address reservation happen before we noticed that the device was a
*hostdev* interface. Those interfaces are hotplugged by just calling
out to qemuDomainAttachHostdevDevice() - that function would then also
attempt to reserve the *same PCI address* that had just been reserved
in qemuDomainAttachNetDevice().

The solution is to move the bit of code that short-circuits out to
virDomainHostdevAttach() up *even earlier* so that no PCI address has
been allocated by the time it's called.

https://bugzilla.redhat.com/show_bug.cgi?id=1744523Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NAndrea Bolognani <abologna@redhat.com>

This introduces semantic validation for SVE-related features,
preventing the user from combining them in invalid ways; it also
automatically enables overall SVE support if any SVE vector
length has been enabled by the user to make sure QEMU behaves
correctly.
Signed-off-by: NAndrea Bolognani <abologna@redhat.com>
Tested-by: NAndrew Jones <drjones@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

The ARM implementation of query-cpu-model-expansion only
supports full expansion, so we have to make sure we're using
that expansion mode if we want to obtain any useful data.
Signed-off-by: NAndrea Bolognani <abologna@redhat.com>
Tested-by: NAndrew Jones <drjones@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

CPU features are available on ARM only wherever the
query-cpu-model-expansion QMP command is available, same as
on s390. Update qemuBuildCpuModelArgStr() to reflect this
fact.
Signed-off-by: NAndrea Bolognani <abologna@redhat.com>
Tested-by: NAndrew Jones <drjones@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Signed-off-by: NAndrea Bolognani <abologna@redhat.com>
Tested-by: NAndrew Jones <drjones@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Mirrors the existing QEMU_CAPS_X86_MAX_CPU.
Signed-off-by: NAndrea Bolognani <abologna@redhat.com>
Tested-by: NAndrew Jones <drjones@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

We're going to use it on non-x86 soon, so it needs a more
generic name: virQEMUCapsObjectPropsMaxCPU.
Signed-off-by: NAndrea Bolognani <abologna@redhat.com>
Tested-by: NAndrew Jones <drjones@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Commit 07552343 added a direct reference to @cookie even though
it may be NULL as shown by a comment a few lines previous - so add
the check here as well.

Found by Coverity
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

Now that we don't have to deal with errors of virBuffer we can also make
this function void.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

The function now does not return an error so we can drop it fully.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Prepare the function for addition of new members to clean.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Use only one switch case selecting job type and decide what's successful
outcome on a case-by-case basis.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Introduce qemuMonitorTransactionBitmapMergeSourceAddBitmap which adds
the appropriate entry into a virJSONValue array to be used with
qemuMonitorTransactionBitmapMerge. Bitmap merging supports two possible
formats and this new helper implements the more universal one specifying
also the source node name.

In addition use the new helper in the testQemuMonitorJSONTransaction
test.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Extract the linking and saving bits of checkpoint creation into
qemuCheckpointCreateFinalize so that qemuCheckpointCreateXML is a bit
simpler and also makes it reusable in the backup code.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Separate out individual steps of creating a checkpoint from
qemuCheckpointCreateXML into separate functions. This makes the function
more readable and understandable and also some of the new functions will
be reusable when we will be creating a checkpoint along with a backup
in the upcoming patches.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Prevent insane configurations by enforcing that disk bitmap for a
checkpoint must match the name of the checkpoint.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

The variable is unused so we can drop it.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

In few places we have the following code pattern:

  int ret;
  ... /* @ret is not accessed here */
  ret = f(...);
  return ret;

This pattern can be written less verbose:

  ...
  return f(...);

This patch was generated with following coccinelle spatch:

  @@
  type T;
  constant C;
  expression f;
  identifier ret;
  @@
  -T ret = C;
   ... when != ret
  -ret = f;
  -return ret;
  +return f;

Afterwards I needed to fix a few places, e.g. comment in
virDomainNetIPParseXML() was removed too because coccinelle
thinks it refers to @ret while in fact it doesn't. Also in few
places it replaced @ret declaration with a few spaces instead of
removing the line. But nothing terribly wrong.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>

These two functions have pattern that's preventing us from
simpler virAsprintf() -> g_strdup_printf() transition. Modify
their logic a bit.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

All OSes that we support have libselinux >= 2.5 except for Ubuntu 16.04
where the version is 2.4.
Signed-off-by: NPavel Hrdina <phrdina@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

So far, the virDomainDeviceFindSCSIController() takes
virDomainDeviceInfo structure which is an overkill. It assumes
that the passed structure is type of
VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE which is not obvious.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

Calling the monitor was convenient for the implementation in
qemuDomainBlockCopyCommon, but causes the snapshot code to call
query-named-block-nodes for every disk.

Fix this by removing the monitor call from
qemuBlockStorageSourceCreateDetectSize so that the data can be reused in
loops.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
ACKed-by: NEric Blake <eblake@redhat.com>

Retrieve data for individual block nodes in a hash table. Currently only
capacity and allocation data is extracted but this will be extended in
the future.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
ACKed-by: NEric Blake <eblake@redhat.com>

In many cases we used virDomainDiskByName to solely look up disk by
target. We have a new helper now so we can replace it.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>

In both replaced cases we have other code that verifies that the bus
can't be changed or that the target is unique, so limiting the search to
disks with same bus makes no sense.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>

In some cases we want to prepare a @src which is not meant to belong to
a disk and thus does not require us to copy the data. Allow passing in
NULL @disk into qemuDomainPrepareDiskSourceData.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>

Note in the comment that this function prepares the storage source based
on the configuration of the disk.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>

The function does not do anything that could fail. Remove the return
value.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>

qemuDomainPrepareDiskSourceData historically prepared everything but
we've split out the majority of the functionality so that it sets up
predominately only according to the configuration of the disk. There
was one leftover bit of setting the gluster debug level from the config.

Split this out into a separate function so that
qemuDomainPrepareDiskSourceData only prepares based on the disk.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
ACKed-by: NEric Blake <eblake@redhat.com>

The @freeTmpPath boolean is used to determine if @tmpPath holds
an allocated memory or is a pointer to a constant string and
therefore if it needs to be freed or not when returning from the
function. Well, we can unify the way we set @tmpPath so that it
always holds an allocated memory and thus always must be freed.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NCole Robinson <crobinso@redhat.com>

There are three cases where vir*DeviceGetPath() returns a const
string. In these cases, the string is initialized in
corresponding vir*DeviceNew() calls which fail if string couldn't
be allocated. There's no point in checking the second time if the
string is NULL.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NCole Robinson <crobinso@redhat.com>

Since its introduction in v1.0.5-rc1-19-g6e13860c the
qemuTeardownHostdevCgroup() does nothing unless the passed
hostdev is a PCI device with VFIO backend. This seems
unnecessary.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NCole Robinson <crobinso@redhat.com>

Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NCole Robinson <crobinso@redhat.com>

There are two types of host devices that require /dev/vfio/vfio
access:

  1) PCI devices with VFIO backend
  2) Mediated devices

Introduce a simple helper that returns true if passed @hostdev
falls in either of the categories.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NCole Robinson <crobinso@redhat.com>

In some places we need to check if a hostdev has VFIO backend.
Because of how complicated virDomainHostdevDef structure is, the
check consists of three lines. Move them to a function and
replace all checks with the function call.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NCole Robinson <crobinso@redhat.com>

Replace all occurrences of
  if (VIR_STRDUP(a, b) < 0)
     /* effectively dead code */
with:
  a = g_strdup(b);
Signed-off-by: NJán Tomko <jtomko@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Replace:
  if (!s && VIR_STRDUP(s, str) < 0)
    goto;
with:
  if (!s)
    s = g_strdup(str);
Signed-off-by: NJán Tomko <jtomko@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

All the callers of these functions only check for a negative
return value.

However, virNetDevOpenvswitchGetVhostuserIfname is documented
as returning 1 for openvswitch interfaces so preserve that.
Signed-off-by: NJán Tomko <jtomko@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Replace all the occurrences of
  ignore_value(VIR_STRDUP_QUIET(a, b));
with
  a = g_strdup(b);
Signed-off-by: NJán Tomko <jtomko@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Replace all the occurrences of
  ignore_value(VIR_STRDUP(a, b));
with
  a = g_strdup(b);
Signed-off-by: NJán Tomko <jtomko@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>