From KVM
(specify "hd,cdrom" boot order -- makes no difference while the target disk is unpartitioned, and is preferred after completing the installation)
(replace stale contents with link to whitepaper)
 
(One intermediate revision by one other user not shown)
Line 1: Line 1:
== Intro ==
+
OVMF "is a project to enable UEFI support for Virtual Machines".
  
OVMF "is a project to enable UEFI support for Virtual Machines". This page tries to give yet another mini-howto about playing with OVMF boot firmware in qemu-kvm virtual machines plus libvirt, deferring heavily to the TianoCore upstream wiki. <em>Do</em> mercilessly edit any inaccuracies or wrong statements.
+
The earlier contents of this article have been replaced with the following link to the OVMF whitepaper:
  
This page is written as of edk2 svn rev 14423 (virtio-blk, virtio-scsi and virtio-net are supported, and several guests can be booted with, and recognize secure boot). For using OVMF directly with the qemu command line, refer to the [https://github.com/tianocore/edk2/blob/master/OvmfPkg/README README]; this page tries to detail OVMF usage under (RHEL-6.4) libvirt.
+
http://www.linux-kvm.org/downloads/lersek/ovmf-whitepaper-c770f8c.txt
  
The recommended way for testing OVMF is installing Gerd Hoffmann's RPM packages from his repo at http://www.kraxel.org/repos/, using yum. (The package to install is <code>edk2.git-ovmf-x64</code>; yum will pull in several dependencies from the repo.)
+
The whitepaper covers everything this article used to cover, and the whitepaper is much more up-to-date and extensive. Please use the History link near the bottom if you'd like to see the previous (now stale) contents.
 
+
The firmware images are located in the <code>/usr/share/edk2.git/ovmf-x64</code> directory. <code>OVMF-pure-efi.fd</code> is a "pure UEFI" image, while <code>OVMF-with-csm.fd</code> includes the Compatibility Support Module (CSM) build of SeaBIOS. <code>bios.bin</code> is a symlink to the latter. Both images support Secure Boot.
+
 
+
== Building from source ==
+
 
+
Clone either the main SVN repository (git svn recommended) or one of the git mirrors listed in the [http://sourceforge.net/apps/mediawiki/tianocore/index.php?title=Source_Control TianoCore wiki].
+
 
+
=== Frequent rebuilds ===
+
 
+
For developers it is recommended to create a branch called <code>base_config</code> or something similar off the master branch (in the git svn or plain git clone), capture the config steps described [http://sourceforge.net/apps/mediawiki/tianocore/index.php?title=How_to_build_OVMF here] in commits (including setting up a reasonable <code>.gitignore</code> file), and keep rebasing <code>base_config</code> after <code>git svn rebase --use-log-author</code> / <code>git pull</code> commands. Fork/rebase your own development branches off/to <code>base_config</code>.
+
 
+
The very first time you build the tree, and after <code>git clean -fdx</code> commands, you must (re)build BaseTools with <code>make -C "$EDK_TOOLS_PATH"</code>. (You're going to have that variable set in your environment after sourcing <code>edksetup.sh</code> in the root project dir; see the [http://sourceforge.net/apps/mediawiki/tianocore/index.php?title=OVMF TianoCore Wiki] again.)
+
 
+
=== One-off builds ===
+
 
+
<code>OvmfPkg/build.sh</code> takes care of BaseTools, configuration (according to command line options) and the main build. One potentially useful option is <code>-n THREADNUMBER</code>, which enables parallel make.
+
 
+
<code>-D FLAG</code> options control optional build features/aspects; among other things, verbosity of OVMF's debug log. Consult <code>OvmfPkg/build.sh</code>, <code>OvmfPkg/README</code>, the <code>OvmfPkg/*.dsc</code> and <code>OvmfPkg/*.fdf</code> files, and Gerd's SRPMs.
+
 
+
== Using the firmware image with libvirt ==
+
 
+
The boot firmware is set in the domain XML file under the <code>/domain/os/loader</code> element.
+
 
+
Neither <code>virt-manager</code> nor <code>virt-install</code> seem to expose this XML node on RHEL-6.4. The following list of commands is one workaround. The <code>EMULATOR</code> bit is discussed later -- it is useful to have a wrapper script in place, between libvirt and qemu, to add custom options.
+
 
+
<pre>
+
# "configuration"
+
NAME=set_guest_name_here
+
INSTALL_ISO=/full/path/to/install/iso
+
DRIVER_ISO=/full/path/to/virtio/driver/iso
+
EMULATOR=/full/path/to/emulator
+
LOADER=/full/path/to/OVMF.fd
+
 
+
# create a domain XML template for guest installation:
+
# - 4 VCPUs, 4G RAM
+
# - virtio target disk, 25 GB in size
+
# - first IDE CD-ROM has install disk
+
# - second IDE CD-ROM has virtio driver disk
+
#  (should only be necessary for proprietary guests without built-in drivers)
+
TARGET_DISK=/var/lib/libvirt/images/"$NAME".img
+
virt-install                                                                \
+
    --connect=qemu:///system                                                \
+
    --name=$NAME                                                            \
+
    --ram=4096                                                              \
+
    --arch=x86_64                                                            \
+
    --machine=rhel6.4.0                                                      \
+
    --vcpus=4                                                                \
+
    --boot=hd,cdrom                                                          \
+
    --disk=path=$TARGET_DISK,size=25,bus=virtio,format=qcow2                \
+
    --disk=path=$INSTALL_ISO,device=cdrom,bus=ide,perms=ro,format=raw        \
+
    --disk=path=$DRIVER_ISO,device=cdrom,bus=ide,perms=ro,format=raw        \
+
    --print-step=1                                                          \
+
| xmlstarlet ed -u /domain/devices/emulator                    -v $EMULATOR \
+
                -s /domain/os                -t elem -n loader -v $LOADER  \
+
    >template.xml
+
 
+
# Import the template to libvirt
+
virsh define template.xml
+
 
+
# Now customize the guest further with "virsh edit" or inside virt-manager,
+
# then start the installation.
+
</pre>
+
 
+
For a virtio-scsi disk, apply the following changes:
+
 
+
<ol>
+
<li>in the <code>TARGET_DISK</code> specification, replace <code>bus=virtio</code> with <code>bus=scsi</code>,</li>
+
<li>append the following options to the <code>xmlstarlet</code> command line:
+
<pre>
+
                -s /domain/devices            -t elem -n controller -v ''          \
+
                -s /domain/devices/controller -t attr -n type      -v scsi        \
+
                -s /domain/devices/controller -t attr -n model      -v virtio-scsi \
+
</pre>
+
</li>
+
</ol>
+
 
+
=== qemu wrapper script under libvirt ===
+
 
+
Libvirt (and its frontends, eg. virsh and virt-manager) provide a convenient way to manage virtual machines. However some qemu command line options are not directly exposed (at least not on a RHEL-6.4 host) that would prove useful otherwise. A script that wraps qemu and plays the emulator role for libvirt allows extra flexibility. On the other side of the coin, it may introduce extra confusion, so use with care.
+
 
+
The full path to the wrapper script is specified in the <code>/domain/devices/emulator</code> element of the libvirt guest XML.
+
 
+
On an SELinux enabled system, the script's context should be set to that of the wrapped emulator binary. See <code>chcon --reference</code>.
+
 
+
(Needless to say, never use a wrapper script in production.)
+
 
+
An example script follows.
+
 
+
<pre>
+
#!/bin/bash
+
set -e -C -u
+
 
+
# Operating modes:
+
# - AD_HOC: use local OVMF & SeaVGABIOS build, ignore iPXE roms,
+
# - AD_HOC_IPXE: same, but make use of ad-hoc iPXE roms,
+
# - KRAXEL_RPMS: use Kraxel's RPMs whole-sale
+
MODE=KRAXEL_RPMS
+
 
+
# Location of ad-hoc ROMs.
+
AD_HOC_PATH=/home/virt-images
+
 
+
# Root installation directory of Kraxel's RPMs.
+
KRAXEL_PATH=/usr/share/edk2.git/ovmf-x64
+
 
+
# Whether to load extra SMBIOS tables.
+
SMBIOS_EXTRA=0
+
 
+
# Argument array constructed for qemu-kvm.
+
NEW_ARGS=()
+
append()
+
{
+
  for I in "$@"; do
+
    NEW_ARGS[${#NEW_ARGS[@]}]=$I
+
  done
+
}
+
 
+
# -vga cirrus found in AD_HOC* modes
+
AD_HOC_CIRRUS=0
+
 
+
# -name XXX found; XXX saved in $NAME
+
NAME=
+
 
+
# previous argument processed
+
LAST=
+
for ARG in "$@"; do
+
  if [ x-vga = x"$LAST" ] && [ cirrus = "$ARG" ] \
+
      && ([ AD_HOC = "$MODE" ] || [ AD_HOC_IPXE = "$MODE" ]); then
+
    AD_HOC_CIRRUS=1
+
    append "$ARG"
+
  elif [ x-device = x"$LAST" ] && [ AD_HOC_IPXE = "$MODE" ]; then
+
    case "$ARG" in
+
      (e1000*)
+
        append "$ARG,romfile=$AD_HOC_PATH/efi-roms/efi-e1000.rom"
+
        ;;
+
      (ne2k_pci*)
+
        append "$ARG,romfile=$AD_HOC_PATH/efi-roms/efi-ne2k_pci.rom"
+
        ;;
+
      (pcnet*)
+
        append "$ARG,romfile=$AD_HOC_PATH/efi-roms/efi-pcnet.rom"
+
        ;;
+
      (rtl8139*)
+
        append "$ARG,romfile=$AD_HOC_PATH/efi-roms/efi-rtl8139.rom"
+
        ;;
+
      (virtio-net-pci*)
+
        append "$ARG,romfile=$AD_HOC_PATH/efi-roms/efi-virtio.rom"
+
        ;;
+
      (*)
+
        append "$ARG"
+
        ;;
+
    esac
+
  elif [ x-bios = x"$LAST" ] && [ KRAXEL_RPMS = "$MODE" ]; then
+
    append "$KRAXEL_PATH/bios.bin" -L "$KRAXEL_PATH"
+
  elif [ x-name = x"$LAST" ]; then
+
    NAME=$ARG
+
    append "$ARG"
+
  else
+
    append "$ARG"
+
  fi
+
 
+
  LAST=$ARG
+
done
+
 
+
if [ -n "$NAME" ]; then
+
  append -debugcon file:/tmp/"$NAME".debug -global isa-debugcon.iobase=0x402 \
+
      -global PIIX4_PM.disable_s3=0 -global PIIX4_PM.disable_s4=0
+
 
+
  if [ 0 -ne $SMBIOS_EXTRA ]; then
+
    append -smbios file=$AD_HOC_PATH/smbios/type3
+
  fi
+
 
+
  if [[ x"$NAME" = xovmf.win2k8* ]] && [ 0 -ne $AD_HOC_CIRRUS ]; then
+
    append -global cirrus-vga.romfile=$AD_HOC_PATH/vgabios-cirrus.csm.bin
+
  fi
+
fi
+
 
+
exec /usr/libexec/qemu-kvm "${NEW_ARGS[@]}"
+
</pre>
+
 
+
This particular script accounts for libvirt invoking the emulator in two forms. XML validation after <code>virsh edit</code> seems to invoke the emulator for verification purposes only, without the <code>-name</code> option. In this case no debug file should be created / rewritten, plus other static options are useless. When libvirt starts the guest, the <code>-name</code> option is present, the script constructs the logfile's name from the corresponding option-argument, and adds some extra static options.
+
 
+
In more detail, the script supports three "operating modes".
+
 
+
<ol>
+
<li>The <code>KRAXEL_RPMS</code> mode is the recommended one for qemu-1.5+. This mode overrides the <code>/domain/os/loader</code> element of the guest XML, and uses the OVMF firmware and any other required files from Gerd's package. Notably, EFI drivers for the emulated / virtio NICs from the iPXE project will be available in the guest as option ROMs.</li>
+
<li>The <code>AD_HOC</code> mode serves development purposes. It disables any iPXE NIC drivers on qemu versions under 1.5 (so that the built-in VirtioNetDxe driver can be tested), lets <code>/domain/os/loader</code>  (= custom OVMF build) take effect, and specifies a custom SeaVGABIOS binary for Windows 2008 R2 guests (based on their name). The Windows 2008 R2 guest is discussed below some more.</li>
+
<li>The <code>AD_HOC_IPXE</code> mode is the same, a development helper, except it gives priority to custom iPXE NIC drivers. (Consult <code>OvmfPkg/README</code> for networking options.)</li>
+
</ol>
+
 
+
The static options enable S3/S4 in RHEL-6.4 qemu, and set the location of OVMF's debug log.
+
 
+
On an SELinux enabled system, a new instance of the same guest will be seclabelled differently from the previous instance, and will fail to overwrite the debug log produced by the previous instance. This could be worked around perhaps by reconfiguring libvirt's labelling practices for the guest, or changing the SELinux profile, or (horribile dictu) flipping SELinux to permissive. Removing the debug file manually before starting the next instance of the guest is simplest.
+
 
+
== Tested guest OS'en ==
+
 
+
<ul>
+
<li>'''Fedora 18''' (XFCE spin tested). When Fedora 18 was released, it co-operated with, and recognized Secure Boot. However some time later the signature verification algorithm changed in TianoCore, and currently OVMF doesn't accept the signatures on Fedora 18 shim. (Fedora 18 works fine on top of OVMF otherwise, when Secure Boot is disabled.)</li>
+
<li>'''Fedora 19''' (XFCE spin tested). It supports Secure Boot on current OVMF. See [https://bugzilla.redhat.com/show_bug.cgi?id=963361#c9 Red Hat Bugzilla 963361 comment 9] and subsequent comments for the setup. In order to transfer certificates from host to guest (for key enrollment), it is recommended to prepare a disk image with a [http://libguestfs.org/ libguestfs] utility, like <code>guestfish</code> or <code>guestmount</code>, and attach it to the virtual machine.</li>
+
<li>'''RHEL-6.3''' and '''RHEL-6.4'''. The <code>grub-efi</code> releases in these RHEL-6 minor releases don't support booting from a virtio-blk disk. Virtio-scsi is supported however (see libvirt XML configuration above). Grub-efi in a future RHEL-6 minor release should hopefully support virtio-blk as well, see [https://bugzilla.redhat.com/show_bug.cgi?id=916016 Red Hat Bugzilla 916016].</li>
+
<li>'''Windows 8'''. In order to test Secure Boot, enroll the keys from [http://blog.hansenpartnership.com/the-microsoft-keys/ James Bottomley's blog], [http://msdn.microsoft.com/en-us/library/windows/desktop/hh847889.aspx start PowerShell as an administrator], then [http://communities.intel.com/community/vproexpert/blog/2012/06/26/microsoft-windows-8--enabling-secure-boot verify Secure Boot].</li>
+
<li>'''Windows Server 2012''', reportedly.</li>
+
<li>'''Windows Server 2008 R2 SP1'''. This guest has a bug (it requires VGA BIOS support even when booted by UEFI firmware). Thankfully, David Woodhouse and Kevin O'Connor implemented a Compatibility Support Module (CSM) for SeaBIOS. Some out-of-tree patches (as of now) are needed for OVMF from http://git.infradead.org/users/dwmw2/edk2.git/; they are included in Gerd's packages and this guest boots fine with them.</li>
+
</ul>
+

Latest revision as of 14:27, 10 March 2015

OVMF "is a project to enable UEFI support for Virtual Machines".

The earlier contents of this article have been replaced with the following link to the OVMF whitepaper:

http://www.linux-kvm.org/downloads/lersek/ovmf-whitepaper-c770f8c.txt

The whitepaper covers everything this article used to cover, and the whitepaper is much more up-to-date and extensive. Please use the History link near the bottom if you'd like to see the previous (now stale) contents.