Add an extra 'inputvol' parameter to the file volume building routines. This
allows us to easily reuse the duplicate functionality.

There is some funkiness here, since we are either dealing with 2 different
pools (which means validation x 2) or the same pool, where we have to be
careful not to deadlock.

We were correctly registering a callback , but weren't telling hal to
actually notify us of any property changes.

The cap_lost and prop_modified callbacks could deadlock if an existing
device needed to be refreshed, since dev_create expects the driver to
be unlocked.

Generating at install time breaks 'make && sudo make install' if root
can't access the users home directory (NFS w/o no_root_squash for example).

Let the underlying driver tell us what it supports. At the moment
we are at least erroneously denying floppy eject/insert.

Don't squash a possibly legitimate libvirtmod error (e.g. some from
clashing libvirt.so versions) with 'Cannot import cygvirtmod'

* src/vbox/vbox_driver.c src/vbox/vbox_tmpl.c: last patch for
  hostonly networks support in vbox driver, patch by Pritesh Kothari
daniel

  patches by Pritesh Kothari
daniel

* docs/schemas/domain.rng src/domain_conf.[ch] src/qemu_driver.c:
  extend the generic code for the RDP and desktop extensions of
  the graphic tag needed for vbox, patch by Pritesh Kothari
Daniel

  patch by Tatsuro Enokura
daniel

daniel

Improves readability, particularly wrt the pending CreateFromXML changes.
This will also help implementing File->Block volume creation in the future,
since some of this code will need to be moved to a backend agnostic file.

vol-clone is a convenience function, requiring only a volume to clone and a
new name.

vol-create-from is a direct mapping to the public API, which allows cloning
across pools, converting between formats, etc, but requires an xml file to be
passed

Also checkin regenerated docs due to API change.

  without an IP address, patch by Ludwig Nussel
Daniel

* src/lxc_container.c: drop LXC CAP_SYS_BOOT capability to avoid
  reboot from inside containers, patch by Ryota Ozaki
Daniel

By checking the pid every retry period, we can quickly determine if
the process crashed at startup, rather than make the user wait for
the entire timeout (3 seconds).