Since a3ab6d42 "apparmor: convert libvirtd profile to a named profile"
the detection of the subelement for qemu_bridge_helper is wrong.

In combination with the older 123cc3e1 "apparmor: allow
/usr/lib/qemu/qemu-bridge-helper" it now detects qemu-bridge-helper no
more with its path, but instead as a proper subelement of the named profile
like: label=libvirtd//qemu_bridge_helper

In the same fashion the reverse rule in the qemu_bridge_helper
sub-profile still uses the path and not the named profile label.

Triggering denies like:
apparmor="DENIED" operation="file_inherit"
  profile="libvirtd//qemu_bridge_helper" pid=5629 comm="qemu-bridge-hel"
  family="unix" sock_type="stream" protocol=0 requested_mask="send receive"
  denied_mask="send receive" addr=none peer_addr=none peer="libvirtd"

This patch fixes the unix socket rules for the communication between
libvirtd and qemu-bridge-helper to match that.

Fixes: a3ab6d42
Fixes: https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1655111Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
Signed-off-by: NChristian Ehrhardt <christian.ehrhardt@canonical.com>

Our nwfilter code doesn't set any timeout on the pcap packet buffer which
means that when DHCP snooping is enabled on a guest interface and
libvirt is trying to learn the IP address from guest's DHCP traffic, it
takes up to 4x longer to ping a guest successfully compared to a case
where nwfilter isn't enabled at all or libvirt uses the cached nwfilter
leases to populate the corresponding rules to ebtables.
With the pcap filter and rate limiting already in place, we should be
able to afford enabling the immediate packet delivery, FWIW immediate
mode was actually the default prior libpcap-1.5.0 (CentOS 6) regardless
of whether a buffer was requested.

The lack of any kind of timeout on the pcap buffer messed with the
libvirt TCK test suite which, even with a generous timeout in place,
timeouts every single time simply because it takes a while until
guest actually starts producing any kind of traffic to fill up
the buffer in place (apart from the DHCP traffic which happens fairly
early on).
Signed-off-by: NErik Skultety <eskultet@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

libpcap-1.5.0 introduced a function to enforce immediate mode (on all
platforms) which the follow-up patches will rely on.
Signed-off-by: NErik Skultety <eskultet@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

Also add a space after the start in some of the cases.
Signed-off-by: NJán Tomko <jtomko@redhat.com>
Reviewed-by: NPeter Krempa <pkrempa@redhat.com>

At the beginning of each profile we have a comment that says when
the profile was last updated. In theory, it makes sense because
one can see immediately if they are using an outdated profile.
However, we don't do a good job in keeping the comments in sync
with reality and also sysadmins should rather use their package
manager to find out libvirt version which installed the profiles.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Acked-by: NChristian Ehrhardt <christian.ehrhardt@canonical.com>

There are two more paths that we are missing in the default
domain profile: /usr/share/edk2-ovmf/ and /usr/share/sgabios/.
These exist on my Gentoo box and contain UEFI and BIOS images
respectively.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Acked-by: NChristian Ehrhardt <christian.ehrhardt@canonical.com>

Even though we construct a domain specific profile for each
domain we start (which should cover domain specific paths), there
is also another file that is included from the profile and which
contains domain agnostic paths (e.g. to cover libraries that qemu
links with). The paths in the file are split into blocks divided
by comments. Sort the paths in each block individually (ignoring
case sensitivity).
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Acked-by: NChristian Ehrhardt <christian.ehrhardt@canonical.com>

With Credit2 being Xen default scheduler, it's definitely the case to
allow Credit2's scheduling parameters to be get and set via libvirt.

This is easy, as Credit and Credit2 have (at least as of now) the very
same parameters ('weight' and 'cap'). So we can just let credit2 pass
the scheduler-type check and the same code will work for both.
Signed-off-by: NDario Faggioli <dfaggioli@suse.com>
Reviewed-by: NJim Fehlig <jfehlig@suse.com>

Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

Aside from itinerant error (actually warning) messages due to an
unrecognized response from qemu, this isn't even necessary - the
migration proceeds successfully to completion anyway.

(I'm not sure where to see this status reported in the API though - do
we need to add an extra state, or recognition of a new event somewhere?)
Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

Normally a PCI hostdev can't be migrated, so
qemuMigrationSrcIsAllowedHostdev() won't permit it. In the case of a a
hostdev network interface that has <teaming type='transient'/> set,
QEMU will automatically unplug the device prior to migration, and
re-plug a corresponding device on the destination. This patch modifies
qemuMigrationSrcIsAllowedHostdev() to allow domains with those devices
to be migrated.
Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

The QEMU driver uses the <teaming type='persistent|transient'
persistent='blah'/> element to setup a "failover" pair of devices -
the persistent device must be a virtio emulated NIC, with the only
extra configuration being the addition of ",failover=on" to the device
commandline, and the transient device must be a hostdev NIC
(<interface type='hostdev'> or <interface type='network'> with a
network that is a pool of SRIOV VFs) where the extra configuration is
the addition of ",failover_pair_id=$aliasOfVirtio" to the device
commandline. These new options are supported in QEMU 4.2.0 and later.

Extra qemu-specific validation is added to ensure that the device
type/model is appropriate and that the qemu binary supports these
commandline options.

The result of this will be:

1) The virtio device presented to the guest will have an extra bit set
in its PCI capabilities indicating that it can be used as a failover
backup device. The virtio guest driver will need to be equipped to do
something with this information - this is included in the Linux
virtio-net driver in kernel 4.18 and above (and also backported to
some older distro kernels). Unfortunately there is no way for libvirt
to learn whether or not the guest driver supports failover - if it
doesn't then the extra PCI capability will be ignored and the guest OS
will just see two independent devices. (NB: the current virtio guest
driver also requires that the MAC addresses of the two NICs match in
order to pair them into a bond).

2) When a migration is requested, QEMu will automatically unplug the
transient/hostdev NIC from the guest on the source host before
starting migration, and automatically re-plug a similar device after
restarting the guest CPUs on the destination host. While the transient
NIC is unplugged, all network traffic will go through the
persistent/virtio device, but when the hostdev NIC is plugged in, it
will get all the traffic. This means that in normal circumstances the
guest gets the performance advantage of vfio-assigned "real hardware"
networking, but it can still be migrated with the only downside being
a performance penalty (due to using an emulated NIC) during the
migration.
Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

The subelement <teaming> of <interface> devices is used to configure a
simple teaming association between two interfaces in a domain. Example:

  <interface type='bridge'>
    <source bridge='br0'/>
    <model type='virtio'/>
    <mac address='00:11:22:33:44:55'/>
    <alias name='ua-backup0'/>
    <teaming type='persistent'/>
  </interface>
  <interface type='hostdev'>
    <source>
      <address type='pci' bus='0x02' slot='0x10' function='0x4'/>
    </source>
    <mac address='00:11:22:33:44:55'/>
    <teaming type='transient' persistent='ua-backup0'/>
  </interface>

The interface with <teaming type='persistent'/> is assumed to always
be present, while the interface with type='transient' may be be
unplugged and later re-plugged; the persistent='blah' attribute (and
in the one currently available implementation, also the matching MAC
addresses) is what associates the two devices with each other. It is
up to the hypervisor and the guest network drivers to determine what
to do with this information.
Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

Presence of the virtio-net-pci option called "failover" indicates
support in a qemu binary of a simplistic bonding of a virtio-net
device with another PCI device. This feature allows migration of
guests that have a network device assigned to a guest with VFIO, by
creating a network bond device in the guest consisting of the
VFIO-assigned device and a virtio-net-pci device, then temporarily
(and automatically) unplugging the VFIO net device prior to migration
(and hotplugging an equivalent device on the migration
destination). (The feature is called "failover" because the bond
device uses the vfio-pci netdev for normal guest networking, but
"fails over" to the virtio-net-pci netdev once the vfio-pci device is
unplugged for migration.)

Full functioning of the feature also requires support in the
virtio-net driver in the guest OS (since that is where the bond device
resides), but if the "failover" commandline option is present for the
virtio-net-pci device in qemu, at least the qemu part of the feature
is available, and libvirt can add the proper options to both the
virtio-net-pci and vfio-pci device commandlines to indicate qemu
should attempt doing the failover during migration.

This patch just adds the qemu capabilities flag "virtio-net.failover".
Signed-off-by: NLaine Stump <laine@redhat.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

* chown: use is conditionally compiled
* configmake: functionality imported to Makefile.am
* getaddrinfo: we have no portability problems
  with Windows impl
* getpass: simplified impl is imported
* mgetgroups: getgrouplist is used directly
* net_if: header includes are conditionalized
* netdb: header includes are conditionalized
* passfd: simplified impl is imported
* posix-shell: functionality was unused & removed
* sigaction: usage is conditionalized
* sigpipe: usage is conditionalized
* stat-time: struct stat is used directly
* strchrnul: usage is eliminated
* strtok_r: usage is not a portability problem
* sys_stat: usage is conditionalized
* uname: rewritten to use native Win32 function to
  get host arch
* waitpid: usage is conditionalized
* wcwidth: rewritten using g_unichar APIs
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The EAI_ADDRFAMILY constant has been removed from FreeBSD
headers, supposedly because it is deprecated by new RFC
drafts.

Previously GNULIB was providing a replacement because
MinGW lacked it too. The replacement provided for MinGW
was thus being used on FreeBSD too, but with a completely
bogus integer value.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

SIGPIPE is not available on the Windows platform.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

There are a large number of different header files that
are related to the sockets APIs. The virsocket.h header
includes all of the relevant headers for Windows and UNIX
in one convenient place. If virsocketaddr.h is already
included, then there's no need for virsocket.h
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

chown and some stat constants are not available on
the Windows platform.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The O_BINARY flag is not defined on all platforms so we must
conditionalize its use once we remove GNULIB.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The O_DIRECT flag is not available on all platforms, so we
must introduce a compat define the same way gnulib does.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The F_DUPFD_CLOEXEC functionality is not available on
some platformms. We must thus explicitly call the
virSetCloexec function once we remove GNULIB's equiv
fix for this.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

Windows uses _O_NOINHERIT as the name for its O_CLOEXEC
equivalent. Define O_CLOEXEC to match this to fix
portability when we remove GNULIB.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The O_DIRECTORY flag causes open() to return an error
if the filename is a directory. There's no obvious
reason why resctrl needs to use this, while the rest of
libvirt code does not. Removing it avoids build issues
on platforms where O_DIRECTORY is not defined, once we
remove GNULIB.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The AC_USE_SYSTEM_EXTENSIONS macro causes things like
"USE_GNU" to be defined, which enables access to OS
specific extensions to POSIX. We currently got this
indirectly via GNULIB's 'extensions' module which is
a dependancy of other GNULIB modules we use.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The S_ISSOCK macro is not available on Windows platforms.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

We currently get the sys/ioctl.h check indirectly
via GNULIB, but this will soon stop happening.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The net/if.h is not portable so we must check for its
existance and avoid using it when missing. Some use
of net/if.h was redundant and could be removed.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

Many of the virProcess APIs are relying on GNULIB providing
POSIX API stubs. Even with these stubs the APIs don't do
anything useful once compiled. We can thus conditionalize
the code so that we don't compile anything at all.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

Cygwin is not a supported build platform for libvirt and
has no testing coverage in our CI systems. Stop pretending
the code is usable and remove it so there is less to port
to Meson.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

A large part of the virCommand code is still built on
WIN32, despite the fact that the core fork() & execve()
functions are not available. So despite succesfully
building most of the code, at runtime the APIs are
none the less unusuable. With the elimination of GNULIB
many of the APIs being used in this code no longer have
portability wrappers/shims for Windows.

Rather than try to add portability wrappers, or do tests
for each individual function, it is clearer to conditionalize
nearly all of the code using #ifdef WIN32.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

polkit is only useful on Linux systems
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

g_mkdir() provides portability to Windows platforms.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The sys/uio.h header is only needed when building logging
code with journald support enabled. Conditionally include
it so that we avoid break on platforms which lack this
header.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The Windows platform does not have the signal handling
support we need, so it must be disabled in several parts
of the codebase.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

This is a simplified variant of gnulib's passfd module
without the portability code that we do not require.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The gnulib stat-time.h header provides wrapper functions
to hide the differences in 'struct stat' between various
platforms.

Linux and FreeBSD support the same names, except for
birthtime which Linux only provides in the new 'statx()'
syscall we're not using. macOS has completely different
naming. Since we only rely on this code in one place
we just use conditionals at time of need.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The virProcess code relies on windows.h and is getting it
indirectly via some GNULIB header fixes. This dependancy
needs to be made explicit.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

The combination of g_unichar_iszerowidth and
g_unichar_iswide is sufficient to replicate the logic
of wcwidth() for libvirt.
Reviewed-by: NPavel Hrdina <phrdina@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>