Added <capabilities> in the <features> section of LXC domains
configuration. This section can contain elements named after the
capabilities like:

  <mknod state="on"/>, keep CAP_MKNOD capability
  <sys_chroot state="off"/> drop CAP_SYS_CHROOT capability

Users can restrict or give more capabilities than the default using
this mechanism.

In the fbd91d49 commit, new scsihostdata dir is added to EXTRA_DIST in
the tests/Makefile.am. However, the directory itself is not created
anywhere, nor in the commit.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Introduce a new function to parse the provided scsi_host parent address
and unique_id value in order to find the /sys/class/scsi_host directory
which will allow a stable SCSI host address

Add a test to scsihosttest to lookup the host# name by using the PCI address
and unique_id value

Add an optional unique_id parameter to nodedev.  Allows for easier lookup
and display of the unique_id value in order to document for use with
scsi_host code.

Introduce a new function to read the current scsi_host entry and return
the value found in the 'unique_id' file.

Add a 'scsihosttest' test (similar to the fchosttest, but incorporating some
of the concepts of the mocked pci test library) in order to read the
unique_id file like would be found in the /sys/class/scsi_host tree.

Between reboots and kernel reloads, the SCSI host number used for SCSI
storage pools may change requiring modification to the storage pool XML
in order to use a specific SCSI host adapter.

This patch introduces the "parentaddr" element and "unique_id" attribute
for the SCSI host adapter in order to uniquely identify the adapter
between reboots and kernel reloads. For now the goal is to only parse
and format the XML. Both will be required to be provided in order to
uniquely identify the desired SCSI host.

The new XML is expected to be as follows:

  <adapter type='scsi_host'>
    <parentaddr unique_id='3'>
      <address domain='0x0000' bus='0x00' slot='0x1f' func='0x2'/>
    </parentaddr>
  </adapter>

where "parentaddr" is the parent device of the SCSI host using the PCI
address on which the device resides and the value from the unique_id file
for the device. Both the PCI address and unique_id values will be used
to traverse the /sys/class/scsi_host/ directories looking at each link
to match the PCI address reformatted to the directory link format where
"domain:bus:slot:function" is found.  Then for each matching directory
the unique_id file for the scsi_host will be used to match the unique_id
value in the xml.

For a PCI address listed above, this will be formatted to "0000:00:1f.2"
and the links in /sys/class/scsi_host will be used to find the host#
to be used for the 'scsi_host' device. Each entry is a link to the
/sys/bus/pci/devices directories, e.g.:

%  ls -al /sys/class/scsi_host/host2
lrwxrwxrwx. 1 root root 0 Jun  1 00:22 /sys/class/scsi_host/host2 -> ../../devices/pci0000:00/0000:00:1f.2/ata3/host2/scsi_host/host2

% cat /sys/class/scsi_host/host2/unique_id
3

The "parentaddr" and "name" attributes are mutually exclusive to identify
the SCSI host number. Use of the "parentaddr" element will be the preferred
mechanism.

This patch only supports to parse and format the XMLs. Later patches will
add code to find out the scsi host number.

Gluster volumes don't start with a leading slash. Our schema for netfs
gluster pools enforces it though. Luckily mount.glusterfs skips it.
Allow a slashless volume name for glusterfs netfs mounts in the schema.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1101999

LXC network devices can now be assigned a custom NIC device name on the
container side. For example, this is configured with:

    <interface type='network'>
      <source network='default'/>
      <guest dev="eth1"/>
    </interface>

In this example the network card will appear as eth1 in the guest.

Test also the TITLE and DESCRIPTION metadata types.

The previous commit 09d4d261 put the interleave at the wrong point;
it didn't allow interleaving with <memory>.

* docs/schema/domaincommon.rng (numatune): Fix interleave location.
* tests/qemuxml2argvdata/qemuxml2argv-numatune-memnode.xml: Adjust test.
Signed-off-by: NEric Blake <eblake@redhat.com>

Currently, we only bind the whole QEMU domain to memory nodes
specified in nodemask altogether.  That, however, doesn't make much
sense when one wants to control from where the memory for particular
guest nodes should be allocated.  QEMU allows us to do that by
specifying 'host-nodes' parameter for the 'memory-backend-ram' object,
so let's use that.
Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

When qemu switched to using OptsVisitor for -numa parameter, it did
two things in the same patch.  One of them is that the numa parameter
is now visible in "query-command-line-options", the second one is that
it enabled using disjoint cpu ranges for -numa specification.  This
will be used in later patch.
Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

That can be lately achieved with by having .param == NULL in the
virQEMUCapsCommandLineProps struct.
Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

There were numerous places where numatune configuration (and thus
domain config as well) was changed in different ways.  On some
places this even resulted in persistent domain definition not to be
stable (it would change with daemon's restart).

In order to uniformly change how numatune config is dealt with, all
the internals are now accessible directly only in numatune_conf.c and
outside this file accessors must be used.
Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

In XML format, by definition, order of fields should not matter, so
order of parsing the elements doesn't affect the end result.  When
specifying guest NUMA cells, we depend only on the order of the 'cell'
elements.  With this patch all older domain XMLs are parsed as before,
but with the 'id' attribute they are parsed and formatted according to
that field.  This will be useful when we have tuning settings for
particular guest NUMA node.
Signed-off-by: NMartin Kletzander <mkletzan@redhat.com>

This patch adds support for the QEMU vhost-user feature to libvirt.
vhost-user enables the communication between a QEMU virtual machine
and other userspace process using the Virtio transport protocol.
It uses a char dev (e.g. Unix socket) for the control plane,
while the data plane based on shared memory.

The XML looks like:

<interface type='vhostuser'>
    <mac address='52:54:00:3b:83:1a'/>
    <source type='unix' path='/tmp/vhost.sock' mode='server'/>
    <model type='virtio'/>
</interface>
Signed-off-by: NMichele Paolino <m.paolino@virtualopensystems.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Add file in storagevolxml2xmlin and storagevolxml2xmlout, let
storagevolxml2xmltest and storagevolschematest cover 'nocow'.
Add test case to storagevolxml2argvtest to cover 'nocow'.
Signed-off-by: NChunyan Liu <cyliu@suse.com>

Assign the value we're comparing:
(val = func()) < 0
instead of assigning the comparison value:
(val = func() < 0)

Both were introduced along with the code,
the TLS tests by commit bd789dff in 0.9.4
net events by commit de87691f in 1.2.2.

Note that the event id type fix is a no-op:
vshNetworkEventIdTypeFromString can only return
-1 (failure) and the event is never used or
0 (the only possible event) and the value of 0 < 0 is still 0.

Rename linuxDomainInterfaceStats to virNetInterfaceStats in order
to allow adding platform specific implementations without
making consumer worrying about specific implementation to be used.

Also, rename util/virstatslinux.c to util/virstats.c so placing
other platform specific implementations into this file don't
look unexpected from the file name.

While preparing to add a capability for active commit, I noticed
that the existing code was abusing int for boolean values.

* src/conf/capabilities.h (_virCapsGuestFeature, _virCapsHost)
(virCapabilitiesNew, virCapabilitiesAddGuestFeature): Improve
types.
* src/conf/capabilities.c (virCapabilitiesNew)
(virCapabilitiesAddGuestFeature): Adjust signature.
* src/bhyve/bhyve_capabilities.c (virBhyveCapsBuild): Update
clients.
* src/esx/esx_driver.c (esxCapsInit): Likewise.
* src/libxl/libxl_conf.c (libxlMakeCapabilities): Likewise.
* src/lxc/lxc_conf.c (virLXCDriverCapsInit): Likewise.
* src/openvz/openvz_conf.c (openvzCapsInit): Likewise.
* src/parallels/parallels_driver.c (parallelsBuildCapabilities):
Likewise.
* src/phyp/phyp_driver.c (phypCapsInit): Likewise.
* src/qemu/qemu_capabilities.c (virQEMUCapsInit)
(virQEMUCapsInitGuestFromBinary): Likewise.
* src/security/virt-aa-helper.c (get_definition): Likewise.
* src/test/test_driver.c (testBuildCapabilities): Likewise.
* src/uml/uml_conf.c (umlCapsInit): Likewise.
* src/vbox/vbox_tmpl.c (vboxCapsInit): Likewise.
* src/vmware/vmware_conf.c (vmwareCapsInit): Likewise.
* src/xen/xen_hypervisor.c (xenHypervisorBuildCapabilities):
Likewise.
* src/xenapi/xenapi_driver.c (getCapsObject): Likewise.
* tests/qemucaps2xmltest.c (testGetCaps): Likewise.
* tests/testutils.c (virTestGenericCapsInit): Likewise.
* tests/testutilslxc.c (testLXCCapsInit): Likewise.
* tests/testutilsqemu.c (testQemuCapsInit): Likewise.
* tests/testutilsxen.c (testXenCapsInit): Likewise.
* tests/vircaps2xmltest.c (buildVirCapabilities): Likewise.
* tests/vircapstest.c (buildNUMATopology): Likewise.
* tests/vmx2xmltest.c (testCapsInit): Likewise.
* tests/xml2vmxtest.c (testCapsInit): Likewise.
Signed-off-by: NEric Blake <eblake@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1113860

We've always done that. Well, until 990e46c4. Point is, if we don't
format model, we may lose a domain on libvirtd restart. If the
seclabel is implicit however, we should skip it's formatting.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1066894

With current code it's possible to have for instance:

virsh dumpxml mydomain | grep seclabel
  <seclabel type='dynamic' model='selinux' relabel='yes'/>
  <seclabel type='dynamic' model='selinux' relabel='yes'/>
  <seclabel type='dynamic' model='selinux' relabel='yes'/>
  <seclabel type='dynamic' model='selinux' relabel='yes'/>
  <seclabel type='dynamic' model='selinux' relabel='yes'/>

what doesn't make any sense. We should reject the XML in the config
parsing phase.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Instead of just returning the parent path, return the complete parent
source structure.

To allow changing the name that is recorded in the overlay of the TOP
image used in a block commit operation, we need to specify the backing
name to qemu. This is done via the "backing-file" attribute to the
block-commit command.

Replace the inline "auth" struct in virStorageSource with a pointer
to a virStorageAuthDefPtr and utilize between the domain_conf, qemu_conf,
and qemu_command sources for finding the auth data for a domain disk

Ressurect the disk-drive-network-iscsi-auth and disk-drive-network-rbd-auth
tests.  Make adjustments to the args and xml file to be compatible with
other changes made to the non "-auth" so that the only difference is the
authentication information.

Adjust the qemuargv2xmltest.c to filter out "<secret" and "</auth>" since
the args -> xml has no concept of usage it doesn't get printed. This results
in the </auth> being printed on the same line as "<secret" and the secret
XML is not closed - a bit of an issue, but soon to be fixed.

Use the probing functionality added in the last patch to turn on
a capability bit when active commit is present, and gate active
commit on that capability.

For my own reference: the difference between BLOCKJOB_SYNC and
BLOCKJOB_ASYNC is whether qemu generated an event at the
conclusion of blockpull; basically, RHEL 6.2 was the only release
of qemu that has the sync semantics and lacks the event.  RHEL
6.3 added blockcopy, but also picked up on the upstream style
of qemu generating events.  As no one is likely to backport
active commit to RHEL 6.2, it's safe for blockcommit to always
require async blockjob support.

Modifying qemucapabilitiestest is painful; the .replies files would
be so much easier if they had comments correlating which command
generated the given reply.  Maybe I'll fix that up later...

* src/qemu/qemu_capabilities.h (QEMU_CAPS_ACTIVE_COMMIT): New
capability.
* src/qemu/qemu_driver.c (qemuDomainBlockCommit): Use the new bit
* src/qemu/qemu_capabilities.c (virQEMUCaps): Name the new bit.
(virQEMUCapsProbeQMPCommands): Set it.
* tests/qemucapabilitiesdata/caps_1.3.1-1.replies: Update.
* tests/qemucapabilitiesdata/caps_1.4.2-1.replies: Likewise.
* tests/qemucapabilitiesdata/caps_1.5.3-1.replies: Likewise.
* tests/qemucapabilitiesdata/caps_1.6.0-1.replies: Likewise.
* tests/qemucapabilitiesdata/caps_1.6.50-1.replies: Likewise.
Signed-off-by: NEric Blake <eblake@redhat.com>

We are about to turn on support for active block commit.  Although
qemu 2.0 was the first version to mostly support it, that version
mis-handles 0-length files, and doesn't have anything available for
easy probing.  But qemu 2.1 fixed bugs, and made life simpler by
letting the 'top' argument be optional.  Unless someone begs for
active commit with qemu 2.0, for now we are just going to enable
it only by probing for qemu 2.1 behavior (anyone backporting active
commit can also backport the optional argument behavior).  This
requires qemu.git commit 7676e2c597000eff3a7233b40cca768b358f9bc9.

Although all our actual uses of block-commit supply arguments for
both base and top, we can omit both arguments and use a bogus
device string to trigger an interesting behavior in qemu.  All QMP
commands first do argument validation, failing with GenericError
if a mandatory argument is missing.  Once that passes, the code
in the specific command gets to do further checking, and the qemu
developers made sure that if device is the only supplied argument,
then the block-commit code will look up the device first, with a
failure of DeviceNotFound, before attempting any further argument
validation (most other validations fail with GenericError).  Thus,
the category of error class can reliably be used to decipher
whether the top argument was optional, which in turn implies a
working active commit.  Since we expect our bogus device string to
trigger an error either way, the code is written to return a
distinct return value without spamming the logs.

* src/qemu/qemu_monitor.h (qemuMonitorSupportsActiveCommit): New
prototype.
* src/qemu/qemu_monitor.c (qemuMonitorSupportsActiveCommit):
Implement it.
* src/qemu/qemu_monitor_json.h (qemuMonitorJSONBlockCommit):
Allow NULL for top and base, for probing purposes.
* src/qemu/qemu_monitor_json.c (qemuMonitorJSONBlockCommit):
Likewise, implementing the probe.
* tests/qemumonitorjsontest.c (mymain): Enable...
(testQemuMonitorJSONqemuMonitorSupportsActiveCommit): ...a new test.
Signed-off-by: NEric Blake <eblake@redhat.com>

The problem is, since 614581f3 domaincapstest is linked with
$(LDADDS) by default. Then, since 94e3f23e the test may be
conditionally linked with $(qemu_LDADDS) which already contains
$(LDADDS). And some linkers doesn't cope with this nicely:

  CCLD     domaincapstest
../src/libvirt_probes.o:(.probes+0x0): multiple definition of `libvirt_event_poll_add_handle_semaphore'
../src/libvirt_probes.o:(.probes+0x0): first defined here
../src/libvirt_probes.o:(.probes+0x2): multiple definition of `libvirt_event_poll_update_handle_semaphore'
../src/libvirt_probes.o:(.probes+0x2): first defined here
../src/libvirt_probes.o:(.probes+0x4): multiple definition of `libvirt_event_poll_remove_handle_semaphore'
../src/libvirt_probes.o:(.probes+0x4): first defined here
../src/libvirt_probes.o:(.probes+0x6): multiple definition of `libvirt_event_poll_dispatch_handle_semaphore'
../src/libvirt_probes.o:(.probes+0x6): first defined here

And so on.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

So far only information on disks and host devices are exposed in the
capabilities XML. Well, at least something. Even a new test is
introduced. The qemu capabilities are stolen from already existing
qemucapabilities test. There's one tricky point though. Functions that
checks host's KVM and VFIO capabilities, are impossible to mock
currently. So in the test, we are setting the capabilities by hand.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Later on, we the qemu capabilities XML parsing code may come handy so
instead of duplicating the code make the already existing one shared.
By the same time, make the function accept file name instead of XML
document stored already in memory.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

This new module holds and formats capabilities for emulator. If you
are about to create a new domain, you may want to know what is the
host or hypervisor capable of. To make sure we don't regress on the
XML, the formatting is not something left for each driver to
implement, rather there's general format function.

The domain capabilities is a lockable object (even though the locking
is not necessary yet) which uses reference counter.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Replace:
if (virBufferError(&buf)) {
    virBufferFreeAndReset(&buf);
    virReportOOMError();
    ...
}

with:
if (virBufferCheckError(&buf) < 0)
    ...

This should not be a functional change (unless some callers
misused the virBuffer APIs - a different error would be reported
then)

So far, we only report an error if formatting the siblings bitmap
in NUMA topology fails.

Be consistent and always report error in virCapabilitiesFormatXML.

This introduces two new attributes "cmd_per_lun" and "max_sectors" same
with the names QEMU uses for virtio-scsi. An example of the XML:

<controller type='scsi' index='0' model='virtio-scsi' cmd_per_lun='50'
max_sectors='512'/>

The corresponding QEMU command line:

-device virtio-scsi-pci,id=scsi0,cmd_per_lun=50,max_sectors=512,
bus=pci.0,addr=0x3
Signed-off-by: NMike Perez <thingee@gmail.com>
Signed-off-by: NJán Tomko <jtomko@redhat.com>

In the test, the snapshot XML is written into a file that's located
under:

  abs_srcdir/vboxsnapshotxmldata/testResult.vbox

However, the abs_srcdir doesn't have to be necessarily writable. It
should have been abs_builddir instead. Moreover, the label in the func
creating the file is called 'fail' while it fulfils the duty of
'cleanup' label.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>