[fedora-virt] Fedora virt status

[Mark McLoughlin]

2009-03-03 Feature freeze (11 days)
  2009-03-10 Beta Freeze (18 days)
  2009-04-14 Final freeze (53 days)

A short report this week, since I was travelling.

libvirt 0.6.0
=============

Dan Berrange fixed the nasty D-Bus threading related segfault
introduced in libvirt 0.6.0:

  https://bugzilla.redhat.com/484553
  http://www.redhat.com/archives/libvir-list/2009-February/msg00357.html

He then pushed a new libvirt-0.6.0 update to fix the last remaining
known issues with the 0.6.0 update in F9/F10:

  * Wed Feb 18 2009 Daniel P. Berrange <berrange at redhat.com> -  0.6.0-4.fc11
  - Fix QEMU startup timeout/race (rhbz #484649)
  - Setup DBus threading. Don't allow dbus to call _exit / change SIGPIPE (rhbz #484553)
  - Fix timeout when autostarting session daemon

sVirt
=====

Dan Walsh posted the latest sVirt patch to libvir-list for review:

  http://www.redhat.com/archives/libvir-list/2009-February/msg00343.html

He also followed that up with a patch to python-virtinst to make it
assign security labels to guests:

  http://www.redhat.com/archives/libvir-list/2009-February/msg00372.html

VT-d
====

David Woodhouse submitted the rwbuf quirk patch to Linus:

  https://lists.linux-foundation.org/pipermail/iommu/2009-February/001117.html

The patch made it into rawhide, closing out:

  https://bugzilla.redhat.com/479996

David also made intel_iommu=on the default again in rawhide:

  * Sat Feb 14 2009 David Woodhouse <David.Woodhouse at intel.com>
  - Enable DMAR by default again now that it works

In the upstream bug report there are reports of a DMAR warning from
the VGA controller on Lenovo x200:

  http://bugzilla.kernel.org/show_bug.cgi?id=12578

   [    0.216043] DMAR: Forcing write-buffer flush capability
   [    0.928007] DMAR:[DMA Write] Request device [00:02.0] fault addr feff5e000
   [    0.928007] DMAR:[fault reason 05] PTE Write access is not set
   00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset
   Integrated Graphics Controller (rev 07)

Bhavesh points out that no further ill effects are noticed after the
warning.

Adar also points out some issues related to ixgbe on Dell Precision
T5400.

PCI Device Assignment
=====================

There was some follow-up to the discussion about device unbinding,
resetting and filtering:

  http://marc.info/?l=kvm&m=123454366317045

Chris Wright posted a patch for "remove_id":

  http://patchwork.kernel.org/patch/7210/

Adar reports more issues with his Dell Precision T5400.

kvmclock
========

The patch to disable kvmclock on machines which don't have
constant_tsc has been merged upstream for 2.6.29 and the patch has
been dropped from rawhide.

QEMU/KVM Merge
==============

Glauber continued his work on BIOS packages:

 https://bugzilla.redhat.com/485418
  Review Request: vgabios - vga option rom for bochs/qemu

  Turns out that this is buildable on all arches with dev86.

 https://bugzilla.redhat.com/485420
  Review Request: openbios - Open Source implementation of IEEE 1275-1994

  Some discussion whether fcode is needed to build, it appears not.

 https://bugzilla.redhat.com/485417
  Review Request: bochs-bios - bios implementation from the bochs project

  Some discussion about whether it would be possible to build on all
  arches with dev86, but it turns out gcc is needed for some bits.

block-rw-range-check.patch
==========================

Eduardo posted a new version of Fedora's only non-upstreamed QEMU
patch, the notorious block-rw-range-check.patch:

  https://bugzilla.redhat.com/486429

An EPEL bug was filed reporting that QEMu snapshots are also broken by
the patch:

  https://bugzilla.redhat.com/486429

Eduardo reports that this is still an issue with his latest patches.

Xen Dom0
========

Michael Young reports on his progress building dom0 enabled kernel
packages

  http://www.redhat.com/archives/fedora-xen/2009-February/msg00031.html

Mass Rebuild
============

A mass rebuild of Fedora 11 is coming soon:

  http://www.redhat.com/archives/fedora-devel-list/2009-February/msg01083.html

These are always the best fun. Expect sparks.

Bugs
====

DOOM-O-METER: 193 open bugs last week, 188 this week. Woohoo!

 https://bugzilla.redhat.com/485564
  Help broken when authenticated through userhelper

  Another virt-manager/consolehelper bug. Our use of consolehelper is
  going away soon, so Cole just closed out the bug and marked it as a
  dup of #459665.

 https://bugzilla.redhat.com/484986
  grub-install fails on virtio disk

 https://bugzilla.redhat.com/480253
  virt-clone fails for clone between storage types

 https://bugzilla.redhat.com/486112
  Qemu should not be using pulseaudio when run as root.

  Some discussion of issues around privileged instances
  of QEMU using pulseaudio and the implications for
  SELinux based confinement of guests.

 https://bugzilla.redhat.com/244787
  incorrect default keymaps

  Some discussion around dropping the -keymap argument
  to QEMU since the guests can now send scancodes rather
  than keysyms.

 https://bugzilla.redhat.com/486575
  Ignores 'target dev=...' attribute in network devices

  libvirt doesn't allow using an explicit vnetXX for a tap device
  name.