nixpkgs/pkgs/build-support/vm/default.nix
Eelco Dolstra 74087918ff * KVM updated to version 74. This one finally has good performance
for userspace networking / Samba again.
* vmtools: use KVM 74 and Linux 2.6.26, and use virtio for networking
  / disk access.

svn path=/nixpkgs/trunk/; revision=12768
2008-08-29 14:08:54 +00:00

823 lines
23 KiB
Nix

{pkgs}:
with pkgs;
rec {
inherit (kernelPackages_2_6_26) kernel klibcShrunk;
modulesClosure = makeModulesClosure {
inherit kernel;
rootModules = ["cifs" "virtio_net" "virtio_pci" "virtio_blk" "virtio_balloon" "nls_utf8"];
};
mountCifs = (makeStaticBinaries stdenv).mkDerivation {
name = "mount.cifs";
src = fetchurl {
name = "mount.cifs.c";
url = "http://websvn.samba.org/cgi-bin/viewcvs.cgi/*checkout*/branches/SAMBA_3_0/source/client/mount.cifs.c?rev=6103";
sha256 = "19205gd3pv8g519hlbjaw559wqgf0h2vkln9xgqaqip2h446qarp";
};
buildInputs = [nukeReferences];
buildCommand = ''
ensureDir $out/bin
gcc -Wall $src -o $out/bin/mount.cifs
strip $out/bin/mount.cifs
nuke-refs $out/bin/mount.cifs
'';
allowedReferences = []; # prevent accidents like glibc being included in the initrd
};
stage1Init = writeScript "vm-run-stage1" ''
#! ${klibcShrunk}/bin/sh.shared -e
echo START
export PATH=${klibcShrunk}/bin:${mountCifs}/bin
mkdir /etc
echo -n > /etc/fstab
mount -t proc none /proc
for o in $(cat /proc/cmdline); do
case $o in
mountDisk=1)
mountDisk=1
;;
command=*)
set -- $(IFS==; echo $o)
command=$2
;;
tmpDir=*)
set -- $(IFS==; echo $o)
export tmpDir=$2
;;
out=*)
set -- $(IFS==; echo $o)
export out=$2
;;
esac
done
for i in $(cat ${modulesClosure}/insmod-list); do
args=
case $i in
*/cifs.ko)
args="CIFSMaxBufSize=4194304"
;;
esac
echo "loading module $i with args $args"
insmod $i $args
done
mount -t tmpfs none /dev
mknod /dev/null c 1 3
mknod /dev/zero c 1 5
mknod /dev/tty c 5 0
mknod /dev/vda b 253 0
ipconfig 10.0.2.15:::::eth0:none
mkdir /fs
if test -z "$mountDisk"; then
mount -t tmpfs none /fs
else
mount -t ext2 /dev/vda /fs
fi
mkdir -p /fs/hostfs
mkdir -p /fs/dev
mount -o bind /dev /fs/dev
mount.cifs //10.0.2.4/qemu /fs/hostfs -o guest,username=nobody
mkdir -p /fs/nix/store
mount -o bind /fs/hostfs/nix/store /fs/nix/store
mkdir -p /fs/tmp
mount -t tmpfs -o "mode=755" none /fs/tmp
mkdir -p /fs/proc
mount -t proc none /fs/proc
mkdir -p /fs/etc
ln -sf /proc/mounts /fs/etc/mtab
echo "Now running: $command"
test -n "$command"
set +e
chroot /fs $command /tmp $out /hostfs/$tmpDir
echo $? > /fs/hostfs/$tmpDir/in-vm-exit
mount -o remount,ro dummy /fs
echo DONE
reboot
'';
initrd = makeInitrd {
contents = [
{ object = stage1Init;
symlink = "/init";
}
];
};
stage2Init = writeScript "vm-run-stage2" ''
#! ${bash}/bin/sh
source $3/saved-env
export NIX_STORE=/nix/store
export NIX_BUILD_TOP="$1"
export TMPDIR="$1"
export PATH=/empty
out="$2"
export ORIG_TMPDIR="$3"
cd "$NIX_BUILD_TOP"
if ! test -e /bin/sh; then
${coreutils}/bin/mkdir -p /bin
${coreutils}/bin/ln -s ${bash}/bin/sh /bin/sh
fi
# For debugging: if this is the second time this image is run,
# then don't start the build again, but instead drop the user into
# an interactive shell.
if test -n "$origBuilder" -a ! -e /.debug; then
${coreutils}/bin/touch /.debug
exec $origBuilder $origArgs
else
export PATH=/bin:/usr/bin:${coreutils}/bin
echo "Starting interactive shell..."
echo "(To run the original builder: \$origBuilder \$origArgs)"
exec ${bash}/bin/sh
fi
'';
qemuCommandLinux = ''
qemu-system-x86_64 \
-nographic -no-reboot \
-net nic,model=virtio -net user -smb / \
-drive file=$diskImage,if=virtio,boot=on \
-kernel ${kernel}/vmlinuz \
-initrd ${initrd}/initrd \
-append "console=ttyS0 panic=1 command=${stage2Init} tmpDir=$TMPDIR out=$out mountDisk=$mountDisk" \
$QEMU_OPTS
'';
vmRunCommand = qemuCommand: writeText "vm-run" ''
export > saved-env
PATH=${coreutils}/bin:${kvm}/bin:${samba}/sbin
diskImage=''${diskImage:-/dev/null}
eval "$preVM"
# Write the command to start the VM to a file so that the user can
# debug inside the VM if the build fails (when Nix is called with
# the -K option to preserve the temporary build directory).
cat > ./run-vm <<EOF
#! ${bash}/bin/sh
diskImage=$diskImage
TMPDIR=$TMPDIR
${qemuCommand}
EOF
chmod +x ./run-vm
source ./run-vm
if ! test -e in-vm-exit; then
echo "Virtual machine didn't produce an exit code."
exit 1
fi
eval "$postVM"
exit $(cat in-vm-exit)
'';
createEmptyImage = {size, fullName}: ''
mkdir $out
diskImage=$out/image
qemu-img create -f qcow $diskImage "${toString size}M"
mkdir $out/nix-support
echo "${fullName}" > $out/nix-support/full-name
'';
createRootFS = ''
mkdir /mnt
${e2fsprogs}/sbin/mke2fs -F /dev/vda
${klibcShrunk}/bin/mount -t ext2 /dev/vda /mnt
if test -e /mnt/.debug; then
exec ${bash}/bin/sh
fi
touch /mnt/.debug
mkdir /mnt/proc /mnt/dev /mnt/sys /mnt/bin
'';
/* Run a derivation in a Linux virtual machine (using Qemu/KVM). By
default, there is no disk image; the root filesystem is a tmpfs,
and /nix/store is shared with the host (via the CIFS protocol to
a Samba instance automatically started by Qemu). Thus, any pure
Nix derivation should run unmodified, e.g. the call
runInLinuxVM patchelf
will build the derivation `patchelf' inside a VM. The attribute
`preVM' can optionally contain a shell command to be evaluated
*before* the VM is started (i.e., on the host). The attribute
`memSize' specifies the memory size of the VM in megabytes,
defaulting to 256. The attribute `diskImage' can optionally
specify a file system image to be attached to /dev/sda. (Note
that currently we expect the image to contain a filesystem, not a
full disk image with a partition table etc.)
If the build fails and Nix is run with the `-K' option, a script
`run-vm' will be left behind in the temporary build directory
that allows you to boot into the VM and debug it interactively. */
runInLinuxVM = attrs: derivation (removeAttrs attrs ["meta" "passthru" "outPath" "drvPath"] // {
builder = "${bash}/bin/sh";
args = ["-e" (vmRunCommand qemuCommandLinux)];
origArgs = attrs.args;
origBuilder = attrs.builder;
QEMU_OPTS = "-m ${toString (if attrs ? memSize then attrs.memSize else 256)}";
});
qemuCommandGeneric = ''
qemu-system-x86_64 \
-nographic -no-reboot \
-smb $(pwd) -hda $diskImage \
$QEMU_OPTS
'';
/* Run a command in a x86 virtual machine image containing an
arbitrary OS. The VM should be configured to do the following:
- Write log output to the serial port.
- Mount //10.0.2.4/qemu via SMB.
- Execute the command "cmd" on the SMB share. It can access the
original derivation attributes in "saved-env" on the share.
- Produce output under "out" on the SMB share.
- Write an exit code to "in-vm-exit" on the SMB share ("0"
meaning success).
- Reboot to shutdown the machine (because Qemu doesn't seem
capable of a APM/ACPI VM shutdown).
*/
runInGenericVM = attrs: derivation (removeAttrs attrs ["meta" "passthru" "outPath" "drvPath"] // {
system = "i686-linux";
builder = "${bash}/bin/sh";
args = ["-e" (vmRunCommand qemuCommandGeneric)];
QEMU_OPTS = "-m ${toString (if attrs ? memSize then attrs.memSize else 256)}";
preVM = ''
diskImage=$(pwd)/image
origImage=${attrs.diskImage}
if test -d "$origImage"; then origImage="$origImage/image"; fi
qemu-img create -b "$origImage" -f qcow $diskImage
echo "$buildCommand" > cmd
eval "$postPreVM"
'';
postVM = ''
cp -prvd out $out
'';
});
/* Like runInLinuxVM, but run the build not using the stdenv from
the Nix store, but using the tools provided by /bin, /usr/bin
etc. from the specified filesystem image, which typically is a
filesystem containing a non-NixOS Linux distribution. */
runInLinuxImage = attrs: runInLinuxVM (attrs // {
mountDisk = true;
/* Mount `image' as the root FS, but use a temporary copy-on-write
image since we don't want to (and can't) write to `image'. */
preVM = ''
diskImage=$(pwd)/image
origImage=${attrs.diskImage}
if test -d "$origImage"; then origImage="$origImage/image"; fi
qemu-img create -b "$origImage" -f qcow $diskImage
'';
/* Inside the VM, run the stdenv setup script normally, but at the
very end set $PATH and $SHELL to the `native' paths for the
distribution inside the VM. */
postHook = ''
PATH=/usr/bin:/bin:/usr/sbin:/sbin
SHELL=/bin/sh
eval "$origPostHook"
'';
origPostHook = if attrs ? postHook then attrs.postHook else "";
/* Don't run Nix-specific build steps like patchelf. */
fixupPhase = "true";
});
/* Create a filesystem image of the specified size and fill it with
a set of RPM packages. */
fillDiskWithRPMs =
{size ? 1024, rpms, name, fullName, preInstall ? "", postInstall ? "", runScripts ? true}:
runInLinuxVM (stdenv.mkDerivation {
inherit name preInstall postInstall rpms;
preVM = createEmptyImage {inherit size fullName;};
buildCommand = ''
${createRootFS}
chroot=$(type -tP chroot)
echo "unpacking RPMs..."
for i in $rpms; do
echo "$i..."
${rpm}/bin/rpm2cpio "$i" | (cd /mnt && ${cpio}/bin/cpio -i --make-directories)
done
eval "$preInstall"
echo "initialising RPM DB..."
PATH=/usr/bin:/bin:/usr/sbin:/sbin $chroot /mnt \
ldconfig -v || true
PATH=/usr/bin:/bin:/usr/sbin:/sbin $chroot /mnt \
rpm --initdb
# Make the Nix store available in /mnt, because that's where the RPMs live.
mkdir -p /mnt/nix/store
${klibcShrunk}/bin/mount -o bind /nix/store /mnt/nix/store
echo "installing RPMs..."
PATH=/usr/bin:/bin:/usr/sbin:/sbin $chroot /mnt \
rpm -iv ${if runScripts then "" else "--noscripts"} $rpms
echo "running post-install script..."
eval "$postInstall"
rm /mnt/.debug
${klibcShrunk}/bin/umount /mnt/nix/store
${klibcShrunk}/bin/umount /mnt
'';
});
/* Generate a script that can be used to run an interactive session
in the given image. */
makeImageTestScript = image: writeScript "image-test" ''
#! ${bash}/bin/sh
if test -z "$1"; then
echo "Syntax: $0 <copy-on-write-temp-file>"
exit 1
fi
diskImage="$1"
if ! test -e "$diskImage"; then
qemu-img create -b ${image}/image -f qcow "$diskImage"
fi
export TMPDIR=$(mktemp -d)
export out=/dummy
export origBuilder=
export origArgs=
export > $TMPDIR/saved-env
mountDisk=1
${qemuCommandLinux}
'';
/* Build RPM packages from the tarball `src' in the Linux
distribution installed in the filesystem `diskImage'. The
tarball must contain an RPM specfile. */
buildRPM = attrs: runInLinuxImage (stdenv.mkDerivation ({
phases = "prepareImagePhase sysInfoPhase buildPhase installPhase";
outDir = "rpms/${attrs.diskImage.name}";
prepareImagePhase = ''
if test -n "$extraRPMs"; then
rpm -iv $extraRPMs
fi
'';
sysInfoPhase = ''
if test -e /etc/fedora-release; then echo "Fedora release: $(cat /etc/fedora-release)"; fi
if test -e /etc/SuSE-release; then echo "SUSE release: $(cat /etc/SuSE-release)"; fi
header "installed RPM packages"
rpm -qa --qf "%{Name}-%{Version}-%{Release} (%{Arch}; %{Distribution}; %{Vendor})\n"
stopNest
'';
buildPhase = ''
eval "$preBuild"
# Hacky: RPM looks for <basename>.spec inside the tarball, so
# strip off the hash.
stripHash "$src"
srcName="$strippedName"
cp "$src" "$srcName" # `ln' doesn't work always work: RPM requires that the file is owned by root
rpmbuild -vv -ta "$srcName" || fail
eval "$postBuild"
'';
installPhase = ''
ensureDir $out/$outDir
find /usr/src -name "*.rpm" -exec cp {} $out/$outDir \;
for i in $out/$outDir/*.rpm; do
header "Generated RPM/SRPM: $i"
rpm -qip $i
stopNest
done
''; # */
} // attrs));
/* Create a filesystem image of the specified size and fill it with
a set of Debian packages. `debs' must be a list of list of
.deb files, namely, the Debian packages grouped together into
strongly connected components. See deb/deb-closure.nix. */
fillDiskWithDebs =
{size ? 1024, debs, name, fullName, postInstall ? null}:
runInLinuxVM (stdenv.mkDerivation {
inherit name postInstall;
debs = (lib.intersperse "|" debs);
preVM = createEmptyImage {inherit size fullName;};
buildCommand = ''
${createRootFS}
echo "initialising Debian DB..."
PATH=$PATH:${dpkg}/bin:${dpkg}/sbin:${glibc}/sbin
# Unpack the .debs. We do this to prevent pre-install scripts
# (which have lots of circular dependencies) from barfing.
echo "unpacking Debs..."
for deb in $debs; do
if test "$deb" != "|"; then
echo "$deb..."
dpkg-deb --extract "$deb" /mnt
fi
done
# Make the Nix store available in /mnt, because that's where the .debs live.
mkdir -p /mnt/inst/nix/store
${klibcShrunk}/bin/mount -o bind /nix/store /mnt/inst/nix/store
${klibcShrunk}/bin/mount -o bind /dev /mnt/dev
# Misc. files/directories assumed by various packages.
touch /mnt/etc/shells
touch /mnt/var/lib/dpkg/status
touch /mnt/var/lib/dpkg/available
touch /mnt/var/lib/dpkg/diversions
# Now install the .debs. This is basically just to register
# them with dpkg and to make their pre/post-install scripts
# run.
echo "installing Debs..."
export DEBIAN_FRONTEND=noninteractive
oldIFS="$IFS"
IFS="|"
for component in $debs; do
IFS="$oldIFS"
echo
echo ">>> INSTALLING COMPONENT: $component"
debs=
for i in $component; do
debs="$debs /inst/$i";
done
chroot=$(type -tP chroot)
PATH=/usr/bin:/bin:/usr/sbin:/sbin $chroot /mnt \
/usr/bin/dpkg --install --force-all $debs < /dev/null
done
echo "running post-install script..."
eval "$postInstall"
rm /mnt/.debug
${klibcShrunk}/bin/umount /mnt/inst/nix/store
${klibcShrunk}/bin/umount /mnt/dev
${klibcShrunk}/bin/umount /mnt
'';
});
/* Generate a Nix expression containing fetchurl calls for the
closure of a set of top-level RPM packages from the
`primary.xml.gz' file of a Fedora or openSUSE distribution. */
rpmClosureGenerator =
{name, packagesList, urlPrefix, packages, archs ? []}:
runCommand "${name}.nix" {buildInputs = [perl perlXMLSimple]; inherit archs;} ''
gunzip < ${packagesList} > ./packages.xml
perl -w ${rpm/rpm-closure.pl} \
./packages.xml ${urlPrefix} ${toString packages} > $out
'';
/* Helper function that combines rpmClosureGenerator and
fillDiskWithRPMs to generate a disk image from a set of package
names. */
makeImageFromRPMDist =
{ name, fullName, size ? 1024, urlPrefix, packagesList, packages
, postInstall ? "", archs ? ["noarch" "i386"], runScripts ? true}:
fillDiskWithRPMs {
inherit name fullName size postInstall runScripts;
rpms = import (rpmClosureGenerator {
inherit name packagesList urlPrefix packages archs;
}) {inherit fetchurl;};
};
/* Like `rpmClosureGenerator', but now for Debian/Ubuntu releases
(i.e. generate a closure from a Packages.bz2 file). */
debClosureGenerator =
{name, packagesList, urlPrefix, packages}:
runCommand "${name}.nix" {} ''
bunzip2 < ${packagesList} > ./Packages
${perl}/bin/perl -I${dpkg} -w ${deb/deb-closure.pl} \
./Packages ${urlPrefix} ${toString packages} > $out
'';
/* Helper function that combines debClosureGenerator and
fillDiskWithDebs to generate a disk image from a set of package
names. */
makeImageFromDebDist =
{name, fullName, size ? 1024, urlPrefix, packagesList, packages, postInstall ? ""}:
fillDiskWithDebs {
inherit name fullName size postInstall;
debs = import (debClosureGenerator {
inherit name packagesList urlPrefix packages;
}) {inherit fetchurl;};
};
/* A bunch of functions that build disk images of various Linux
distributions, given a set of top-level package names to be
installed in the image. */
diskImageFuns = {
fedora2i386 = args: makeImageFromRPMDist ({
name = "fedora-core-2-i386";
fullName = "Fedora Core 2 (i386)";
packagesList = fetchurl {
url = mirror://fedora/linux/core/2/i386/os/repodata/primary.xml.gz;
sha256 = "1nq1k2k0nzkii737cka301f0vbd2ix2wsfvi6bblpi748q6h2w4k";
};
urlPrefix = mirror://fedora/linux/core/2/i386/os;
runScripts = false;
} // args);
fedora3i386 = args: makeImageFromRPMDist ({
name = "fedora-core-3-i386";
fullName = "Fedora Core 3 (i386)";
packagesList = fetchurl {
url = mirror://fedora/linux/core/3/i386/os/repodata/primary.xml.gz;
sha256 = "13znspn4g1bkjkk47393k9chswgzl6nx1n0q6h2wrw52c7d9nw9i";
};
urlPrefix = mirror://fedora/linux/core/3/i386/os;
archs = ["noarch" "i386" "i586"];
runScripts = false;
} // args);
fedora5i386 = args: makeImageFromRPMDist ({
name = "fedora-core-5-i386";
fullName = "Fedora Core 5 (i386)";
packagesList = fetchurl {
url = mirror://fedora/linux/core/5/i386/os/repodata/primary.xml.gz;
sha256 = "0lfk4mzrpiyls8h7k9ckc3vgywbmg05zsr4ag6qakgnv9gljijig";
};
urlPrefix = mirror://fedora/linux/core/5/i386/os;
} // args);
fedora7i386 = args: makeImageFromRPMDist ({
name = "fedora-7-i386";
fullName = "Fedora 7 (i386)";
packagesList = fetchurl {
url = mirror://fedora/linux/releases/7/Fedora/i386/os/repodata/primary.xml.gz;
sha256 = "0zq7ifirj45wry7b2qkm12qhzzazal3hn610h5kwbrfr2xavs882";
};
urlPrefix = mirror://fedora/linux/releases/7/Fedora/i386/os;
} // args);
fedora8i386 = args: makeImageFromRPMDist ({
name = "fedora-8-i386";
fullName = "Fedora 8 (i386)";
packagesList = fetchurl {
url = mirror://fedora/linux/releases/8/Fedora/i386/os/repodata/primary.xml.gz;
sha256 = "0vr9345rrk0vhs4pc9cjp8npdkqz0xqyirv84vhyfn533m9ws36f";
};
urlPrefix = mirror://fedora/linux/releases/8/Fedora/i386/os;
} // args);
fedora9i386 = args: makeImageFromRPMDist ({
name = "fedora-9-i386";
fullName = "Fedora 9 (i386)";
packagesList = fetchurl {
url = mirror://fedora/linux/releases/9/Fedora/i386/os/repodata/primary.xml.gz;
sha256 = "18780xgyag5acx79warcpvzlfkm0mni8xawl6jjvgxg9n3lp6zg0";
};
urlPrefix = mirror://fedora/linux/releases/9/Fedora/i386/os;
} // args);
opensuse103i386 = args: makeImageFromRPMDist ({
name = "opensuse-10.3-i586";
fullName = "openSUSE 10.3 (i586)";
packagesList = fetchurl {
url = mirror://opensuse/distribution/10.3/repo/oss/suse/repodata/primary.xml.gz;
sha256 = "0zb5kxsb755nqq9i8jdclmanacyf551ncx6a011v9jqphsvyfvd7";
};
urlPrefix = mirror://opensuse/distribution/10.3/repo/oss/suse/;
archs = ["noarch" "i586"];
} // args);
ubuntu710i386 = args: makeImageFromDebDist ({
name = "ubuntu-7.10-gutsy-i386";
fullName = "Ubuntu 7.10 Gutsy (i386)";
packagesList = fetchurl {
url = mirror://ubuntu/dists/gutsy/main/binary-i386/Packages.bz2;
sha1 = "8b52ee3d417700e2b2ee951517fa25a8792cabfd";
};
urlPrefix = mirror://ubuntu;
} // args);
debian40r3i386 = args: makeImageFromDebDist ({
name = "debian-4.0r3-etch-i386";
fullName = "Debian 4.0r3 Etch (i386)";
packagesList = fetchurl {
url = mirror://debian/dists/etch/main/binary-i386/Packages.bz2;
sha256 = "7a8f2777315d71fd7321d1076b3bf5f76afe179fe66c2ce8e1ff4baed6424340";
};
urlPrefix = mirror://debian;
} // args);
};
/* Common packages for Fedora images. */
commonFedoraPackages = [
"autoconf"
"automake"
"basesystem"
"bzip2"
"curl"
"diffutils"
"fedora-release"
"findutils"
"gawk"
"gcc-c++"
"gzip"
"make"
"patch"
"perl"
"pkgconfig"
"rpm"
"rpm-build"
"tar"
"unzip"
];
/* Common packages for openSUSE images. */
commonOpenSUSEPackages = [
"aaa_base"
"autoconf"
"automake"
"bzip2"
"curl"
"devs"
"diffutils"
"findutils"
"gawk"
"gcc-c++"
"gzip"
"make"
"patch"
"perl"
"pkg-config"
"rpm"
"tar"
"unzip"
"util-linux"
];
/* Common packages for Debian/Ubuntu images. */
commonDebianPackages = [
"base-passwd"
"dpkg"
"libc6-dev"
"perl"
"sysvinit"
"bash"
"gzip"
"bzip2"
"tar"
"grep"
"findutils"
"g++"
"make"
"curl"
"patch"
"diff"
];
/* A bunch of disk images. */
diskImages = {
redhat9i386 = fillDiskWithRPMs {
name = "redhat-9-i386";
fullName = "Red Hat Linux 9 (i386)";
size = 1024;
rpms = import ./rpm/redhat-9-i386.nix {inherit fetchurl;};
};
suse90i386 = fillDiskWithRPMs {
name = "suse-9.0-i386";
fullName = "SUSE Linux 9.0 (i386)";
size = 1024;
rpms = import ./rpm/suse-9-i386.nix {inherit fetchurl;};
# Urgh. The /etc/group entries are installed by aaa_base (or
# something) but due to dependency ordering, that package isn't
# installed yet by the time some other packages refer to these
# entries.
preInstall = ''
echo 'bin:x:1:daemon' >> /mnt/etc/group
echo 'tty:x:5:' >> /mnt/etc/group
echo 'disk:x:6:' >> /mnt/etc/group
echo 'lp:x:7:' >> /mnt/etc/group
echo 'uucp:x:14:' >> /mnt/etc/group
echo 'audio:x:17:' >> /mnt/etc/group
echo 'video:x:33:' >> /mnt/etc/group
'';
};
fedora2i386 = diskImageFuns.fedora2i386 { packages = commonFedoraPackages; };
fedora3i386 = diskImageFuns.fedora3i386 { packages = commonFedoraPackages; };
fedora5i386 = diskImageFuns.fedora5i386 { packages = commonFedoraPackages; };
fedora7i386 = diskImageFuns.fedora7i386 { packages = commonFedoraPackages; };
fedora8i386 = diskImageFuns.fedora8i386 { packages = commonFedoraPackages; };
fedora9i386 = diskImageFuns.fedora9i386 { packages = commonFedoraPackages; };
opensuse103i386 = diskImageFuns.opensuse103i386 { packages = commonOpenSUSEPackages; };
ubuntu710i386 = diskImageFuns.ubuntu710i386 { packages = commonDebianPackages; };
debian40r3i386 = diskImageFuns.debian40r3i386 { packages = commonDebianPackages; };
};
}