- enable rbd and spice modules (needs external libraries)

- Make the kvm-amd / kvm-intel modules load automatically instead of using
  the init script.  Reportedly kvm modules in recent kernels can be loaded
  even if other virtualization technologies, such as vmware, are being used
  (previously the two sets of kernel modules were conflicting with each other).
  This needs to be verified.

- /etc/init.d/qemu-kvm script might replace eth0 with a switch sw0.
  Config questions to be asked, possibly alternative default network
  setups supplied etc.
  Note that this is quite fragile, especially in case of more
  comples network settings.  Maybe create a script that updates
  /etc/network/interfaces instead.
  Mjt: I don't like the whole idea of messing up with networking like this...

- In /etc/kvm/kvm-ifup, be more flexible.  Also don't depend on bridge-utils.
  (already a bit more flexible than before, maby that's enough...)

- The whole networking should probably be left to various management tools
  instead, but kvm is useful without such tools.  At least provide some examples,
  especially how to use networking when running as non-root.
  Note that many howtos for the subject suggests using sudo to run ifconfig,
  brctl and the like, -- this is wrong approach, you have to grant whole
  network configuration to the user in question, but he only need to access
  one script as root instead of being able to mess with whole networking
  on the system

- Why we suggest samba?  If it is only for user-mode networking, is it even
  worth to mention?  User networking is too slow to be useful.

- Create a kvm-doc package from the texinfo source of the full manual in
  qemu/qemu-doc.texi.  The kvm upstream for qemu-doc.texi isn't up to date with
  the qemu upstream on the manual, though.  Which also needs to be verified if
  it's still the case with qemu-kvm packages (should be pretty up-to-date).
  Mjt is not sure if it's worth to create such package when we have the
  manpages (generated from the same .texi files).

- Review and patch the documentation, look for proper naming.  Right now it
  all refers to qemu or qemu-doc, so basically s/qemu/kvm/g.  Probably a sort
  of a sed script during install.  Also mention there what default network
  script does (see above about networking too)

- Provide some missing manpages: kvm_stat kvm-io

- Create a qemu-kvm-dev package from the files that are currently removed in rules

- Provide packages for non-x86 arches.  Powerpc, s390.  Need hardware to test.

- kvm/libfdt can't be cleaned up if no configure were run.  Used on ppc.
  Note that now 'clean' target isn't being executed if no configure step
  were performed.

- Review SASL support and qemu.sasl in particular.  Sasl is used for vnc.
  Need to ensure we're using the right name in the source too.

- for !x86 arches it might be necessary to include lintian-overrides file
  to list things like:
   qemu-kvm: arch-dependent-file-in-usr-share ./usr/share/kvm/openbios-ppc
   qemu-kvm: binary-from-other-architecture ./usr/share/kvm/openbios-ppc
   qemu-kvm: statically-linked-binary ./usr/share/kvm/openbios-ppc
