While running make check, I noticed that it was actually using
the virsh binary from my system, in /usr/bin/virsh, and not the
one that was just compiled.  This is actually caused by a bug
in Makefile.am, where we didn't update the PATH to include tools.

While here, I also updated all of the scripts to properly define
the srcdir, abs_top_srcdir, and abs_top_builddir environment
variables.  This is required if you want to be able to run the
tests standalone (i.e. ./test instead of from make check).  I've
tested this on both RHEL-5 and Fedora-10 machines, and make check
works on both, as does running the individual tests by hand.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

Otherwise logrotate barfs:

  error: error accessing /var/log/libvirt/uml: No such file or directory
  error: libvirtd:1 glob failed for /var/log/libvirt/uml/*.log
  error: found error in /var/log/libvirt/qemu/*.log /var/log/libvirt/uml/*.log /var/log/libvirt/lxc/*.log , skipping

* qemud/Makefile.am: always create /var/log/libvirt/{lxc,uml} when
  installing the logrotate conf; not ideal, but easier than making
  the logrotate conf depend on which drivers are enabled

The devhelp/ directory files depend on libvirt-api.xml being
uptodate, but automake always processes SUBDIRS before the
current directory. So devhelp would be built and then the
libvirt-api.xml re-build, invalidating the devhelp data again.

To fix this all the rules for devhelp are moved directly
into the docs/Makefile.am allowing make to see the global
dependancy chain and thus build things in the correct order

* docs/Makefile.am: Add rules for devhelp rebuild
* docs/devhelp/Makefile.am: Remove devhelp rebuild
* configure.in: Remove docs/devhelp/Makefile.am

* docs/apibuild.py: Make uniq() function sort keys before returning
  them to ensure consitent ordering of <exports> in XML file

Also add some XML examples.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

Add virNodeDeviceParseFile, and make virNodeDeviceParseNode non-static. These
will be used by the test driver.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

Add a simple 'computer' device for the default driver. Only implement
the basic calls, no creation or destroy happening.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

Will be used by test driver node device implementation.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

Signed-off-by: NCole Robinson <crobinso@redhat.com>

Throw error in GetBridgeName if net has no bridge.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

In the generated bindings, custom classes are squashed against the following
class, which hurts readability.

A mistake in the generator was causing virInterface methods to be generated
with unpredicatable names ('ceUndefine', instead of just 'undefine'). This
fixes the method names to match existing convention.

Does anyone care if we are breaking API compat? My guess is that no one is
using the python interface bindings yet.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

The API docs explictly warn that we shouldn't use the C vir*GetConnect calls
in bindings: doing so can close the internal connection pointer and cause
things to get screwy. Implement these calls in python.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

They are only for use in implementing the bindings, so shouldn't be
exposed to regular API users.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

A special case in the generator wasn't doing its job, and duplicate
conn.createXML functions were being generated. The bindings diff is:

@@ -1079,14 +1079,6 @@ class virConnect:
         return __tmp

     def createXML(self, xmlDesc, flags):
-        """Create a new device on the VM host machine, for example,
-           virtual HBAs created using vport_create. """
-        ret = libvirtmod.virNodeDeviceCreateXML(self._o, xmlDesc, flags)
-        if ret is None:raise libvirtError('virNodeDeviceCreateXML() failed', conn=self)
-        __tmp = virNodeDevice(self, _obj=ret)
-        return __tmp
-
-    def createXML(self, xmlDesc, flags):
         """Launch a new guest domain, based on an XML description
           similar to the one returned by virDomainGetXMLDesc() This
           function may requires privileged access to the hypervisor.
@@ -1327,6 +1319,14 @@ class virConnect:
         __tmp = virNetwork(self, _obj=ret)
         return __tmp

+    def nodeDeviceCreateXML(self, xmlDesc, flags):
+        """Create a new device on the VM host machine, for example,
+           virtual HBAs created using vport_create. """
+        ret = libvirtmod.virNodeDeviceCreateXML(self._o, xmlDesc, flags)
+        if ret is None:raise libvirtError('virNodeDeviceCreateXML() failed', conn=self)
+        __tmp = virNodeDevice(self, _obj=ret)
+        return __tmp
+
     def nodeDeviceLookupByName(self, name):
         """Lookup a node device by its name. """
         ret = libvirtmod.virNodeDeviceLookupByName(self._o, name)
Signed-off-by: NCole Robinson <crobinso@redhat.com>

xmllib has been deprecated since python 2.0, and running the generator throws
a warning. Move to using xml.sax
Signed-off-by: NCole Robinson <crobinso@redhat.com>

FastParser uses sgmlop, a non-standard python module meant as a replacement
for xmllib (which is deprecated since python 2.0). Fedora doesn't even carry
this module, and the generator doesn't have high performance requirements, so
just rip the code out.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

--with-python currently already works for enabling/disabling the python
bindings, but doesn't show up in the help output.
Signed-off-by: NCole Robinson <crobinso@redhat.com>

* daemon/libvirtd.logrotate.in: change to weekly rotation of logs,
  keep a month worth of data and also extend to cover LXC and UML
  domain logs

* docs/Makefile.am: Fix syntax error in NEWS rule, '$' should be '$$'
  to escape correctly

The code which updated the message length after writing the
payload wrote the updated length word in the wrong place since
the XDR object was given a buffer pointing to the start of the
header payload, rather than message start.

* daemon/remote.c: Fix updating of event message length so that
  we actually send the payload, not just the header

Fix "make rpcgen", broken by the directory reorganization.

* src/Makefile.am (rpcgen): Fix path to rpcgen_fix.pl.

Fix migration, broken in two different ways by the QEMU monitor
abstraction.  Note that the QEMU console emits a "\r\n" as the
line-ending.

* src/qemu/qemu_monitor_text.c (qemuMonitorGetMigrationStatus):
Fix "info migrate" command and its output's parsing.

Implementation of tunnelled migration, using a Unix Domain Socket
on the qemu backend.  Note that this requires very new versions of
qemu (0.10.7 at least) in order to get the appropriate bugfixes.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

* cfg.mk: use $(srcdir)/ prefix for Makefile.nonreentrant include
* examples/domain-events/events-c/Makefile.am tools/Makefile.am
  examples/hellolibvirt/Makefile.am: extend the include paths to
  use $(top_srcdir)/include too.

Fix a few mispellings :-) of "successfully" and regenerate
docs/libvirt-*.xml.
* src/libvirt.c: Fix typos.
* src/secret/secret_driver.c: Fix typos.
* docs/libvirt-api.xml: Regenerate.
* docs/libvirt-refs.xml: Regenerate.

The upcoming tunnelled migration needs to be able to set
a migration in progress in the background, as well as
be able to cancel a migration when a problem has happened.
This patch allows for both of these to properly work.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

A simple misplaced break out of a switch results in:

  libvir: error : Failed to open file '/sys/bus/pci/devices/0000:00:54c./vendor': No such file or directory
  libvir: error : Failed to open file '/sys/bus/pci/devices/0000:00:54c./device': No such file or directory
  libvir: error : this function is not supported by the hypervisor: Failed to read product/vendor ID for 0000:00:54c.

when trying to passthrough a USB host device to qemu.

* src/security_selinux.c: fix a switch/break thinko

* src/conf/domain_conf.c: a simple typo in an XML domain file could lead
  to a crash, because we called STRPREFIX() on the looked up value without
  checking it was non-null.

* src/conf/domain_conf.c: when declaring a <interface type="bridge">
  tag, <source> needs a "bridge" attribute, but the parser complains
  about a missing "dev" attribute.

* Makefile.am: examples/domain-events/events-python should be added
  to dist tarball
* libvirt.spec.in: there is no makefile in domain-events but in
  domain-events/events-c and python/libvirtclass.txt has vanished

* docs/schemas/domain.rng: allow one <description> tag in the top level
  of the <domain> to store user information as text
* src/conf/domain_conf.c src/conf/domain_conf.h: extend the structure
  to store this text, grab it at parse time and save it back when
  present after <uuid>

Signed-off-by: NChris Lalancette <clalance@redhat.com>

Signed-off-by: NChris Lalancette <clalance@redhat.com>

This is necessary for the dispatch functions to be able to use
streams in the future.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

Right now, the stream stuff assumes that a stream is always
going to be used for transmit.  This is not the case, and in
fact doesn't work with the tunnelled migration stuff.  Add
a flag to remoteClientStream() to allow it to do RX only.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

Signed-off-by: NChris Lalancette <clalance@redhat.com>

Signed-off-by: NChris Lalancette <clalance@redhat.com>

Function comments for virStreamEvent{Add,Update,Remove}Callback() are
missing a trailing ':'. Therefore apibuild.py fails to parse the comment
and warns about the missing ':'.

* docs/libvirt-api.xml, docs/libvirt-refs.xml: updated by apibuild.py
* src/libvirt.c: add missing ':' in function comments

The actual type of size_t is architecture dependent. Because the len
parameter is used as unsigned int in remoteSendStreamData(), change its
type to unsigned int.

* daemon/dispatch.[ch]: change size_t to unsigned int for
  remoteSendStreamData()