[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] /etc/grub.d/09-xen for generating grub.cfg for hypervisor boot entries.
Yes that should do it and no corresponding boot/config-$(uname -r) means nothing will be generated for the respective kernel. There could also be a check for CONFIG_XEN=y to support Xenlinux kernels. That entry should not exist in pvops .config so a special menu entry tag could also be generated depicting kernel type if desirable. Those are things I might add in afterwards if you are not interested in them. Bruce Edge wrote: Perhaps grepping for CONFIG_XEN_DOM0=y in the boot/config-XX would validate the dom0 kernel compatibility better then looking for -xen in the name.-Bruce -BruceOn Wed, May 19, 2010 at 10:07 AM, Richie <listmail@xxxxxxxxxxxx <mailto:listmail@xxxxxxxxxxxx>> wrote:Perhaps it goes without saying, but I don't think my suggestion for a xen in the image name check will be an accurate solution for someone that does not roll their own kernels. I always know that my kernels have xen in the name when they are pvops/dom0 capable and sxen when they are Xenlinux. Also, technically, for Xenlinux kernels, grub should not be generating bare metal boot entries. If anything, this could also be used as a basis for a script that generates an initial 40_custom file. That can then be hand edited before running update-grub. Bruce Edge wrote: Good points all. Will incorporate, retest and repost the resultant script. Thanks -Bruce On Tue, May 18, 2010 at 5:24 PM, Richie <listmail@xxxxxxxxxxxx <mailto:listmail@xxxxxxxxxxxx> <mailto:listmail@xxxxxxxxxxxx <mailto:listmail@xxxxxxxxxxxx>>> wrote: I like the idea myself and I haven't seen anything in the wiki's other than manual creation steps. Just an opinion here, but why not use "dummy=dummy" as opposed to first parameter duplication? My understanding is that dummy is used to avoid this same bug. Either way we it avoids having to hardcode root= into the kernel cmdline .config parameter :) I don't know if parameter duplication would break things when the bug is fixed or not, but the dummy parameter shouldn't. I also think it might be viewed as something more familiar, perhaps self explanatory, whereas the parameter duplication may cause confusion. I skimmed the code and have not tested it. I don't see that it is specifically trying to ensure that the kernel is Xenlinux or pvops... Not that I know of a proper way to do such or if its even pratical. Aren't most kernels now pvops (thus bootable under xen) but not necessarily dom0 capable? I think a spin on this would be if one wanted to limit the Xen entries to kernels with "xen" (ie. --append-to-version) in the name. Perhaps the code would change as follows? <snip> list=`for i in /boot/vmlinu[xz]-*xen* /vmlinu[xz]-*xen* ; do if grub_file_is_not_garbage "$i" ; then echo -n "$i " ; fi done` <snip> Bruce Edge wrote: If this has already been done, please forgive me. However, if not, I'd like to submit this as a mechanism for generating a bootable grub2 stanza for hypervisors. As the /etc/grub.d/* files rely on defaults in /etc/default/grub, I added the following Xen specific variable: GRUB_CMDLINE_XEN_DEFAULT="console=com1 115200,8n1 dom0_mem=512M dom0_max_vcpus=1 dom0_vcpus_pin=true iommu=1,passthrough,no-intremap loglvl=all loglvl_guest=all loglevl=10 debug acpi=force apic=on apic_verbosity=verbose numa=on" The script itself is a hacked version of the10-linux that comes with grub2. This is 09-xen so it places it's boot entries ahead of the non-xen entries. The resulting grub.cfg entry looks like: ### BEGIN /etc/grub.d/09_xen ### insmod lvm set root=(system-dom0_0) menuentry "Xen osa-dom0 6.0.13-05, linux 2.6.32.12" { multiboot /boot/xen.gz /boot/xen.gz console=com1 115200,8n1 dom0_mem=512M dom0_max_vcpus=1 dom0_vcpus_pin=true iommu=1,passthrough,no-intremap loglvl=all loglvl_guest=all loglevl=10 debug acpi=force apic=on apic_verbosity=verbose numa=on module /boot/vmlinuz-2.6.32.12 /boot/vmlinuz-2.6.32.12 root=UUID=a3764d7d-6292-4f08-8ece-480e54c77229 ro earlyprintk=xen loglevel=10 debug acpi=force console=hvc0,115200n8 module /boot/initrd.img-2.6.32.12 /boot/initrd.img-2.6.32.12 } ### END /etc/grub.d/09_xen ### Note the duplication of the first params. I believe there's a bug that drops the 1st param so this could be changed later. #! /bin/sh -e prefix=/usr exec_prefix=${prefix} libdir=${exec_prefix}/lib . ${libdir}/grub/update-grub_lib if [ "x${GRUB_DISTRIBUTOR}" = "x" ] ; then OS=GNU/Linux else OS="${GRUB_DISTRIBUTOR}" fi # Source grub defaults . /etc/default/grub # loop-AES arranges things so that /dev/loop/X can be our root device, but # the initrds that Linux uses don't like that. case ${GRUB_DEVICE} in /dev/loop/*|/dev/loop[0-9]) GRUB_DEVICE=`losetup ${GRUB_DEVICE} | sed -e "s/^[^(]*(\([^)]\+\)).*/\1/"` ;; esac if [ "x${GRUB_DEVICE_UUID}" = "x" ] || [ "x${GRUB_DISABLE_LINUX_UUID}" = "xtrue" ] \ || ! test -e "/dev/disk/by-uuid/${GRUB_DEVICE_UUID}" ; then LINUX_ROOT_DEVICE=${GRUB_DEVICE} else LINUX_ROOT_DEVICE=UUID=${GRUB_DEVICE_UUID} fi test_gt () { local a=`echo $1 | sed -e"s,.*/vmlinu[zx]-,,g;s/[._-]\(pre\|rc\|test\|git\|old\)/~\1/g"`local b=`echo $2 | sed -e"s,.*/vmlinu[zx]-,,g;s/[._-]\(pre\|rc\|test\|git\|old\)/~\1/g"`if [ "x$b" = "x" ] ; then return 0 fi dpkg --compare-versions "$a" gt "$b" return $? } find_latest () { local a="" for i in $@ ; do if test_gt "$i" "$a" ; then a="$i" fi done echo "$a" } list=`for i in /boot/vmlinu[xz]-* /vmlinu[xz]-* ; do if grub_file_is_not_garbage "$i" ; then echo -n "$i " ; fi done` while [ "x$list" != "x" ] ; do linux=`find_latest $list` echo "Found linux image: $linux" >&2 basename=`basename $linux` dirname=`dirname $linux` rel_dirname=`make_system_path_relative_to_its_root $dirname` version=`echo $basename | sed -e "s,^[^0-9]*-,,g"` alt_version=`echo $version | sed -e "s,\.old$,,g"` linux_root_device_thisversion="${LINUX_ROOT_DEVICE}" initrd= for i in "initrd.img-${version}" "initrd-${version}.img" \ "initrd.img-${alt_version}" "initrd-${alt_version}.img"; do if test -e "${dirname}/${i}" ; then initrd="$i" break fi done if test -n "${initrd}" ; then echo "Found initrd image: ${dirname}/${initrd}" >&2 else # "UUID=" magic is parsed by initrds. Since there's no initrd, it can't work here. linux_root_device_thisversion=${GRUB_DEVICE} fi cat << EOF insmod lvm set root=(system-dom0_0) menuentry "Xen ${OS}, linux ${version}" { multiboot /boot/xen.gz /boot/xen.gz $GRUB_CMDLINE_XEN_DEFAULT module ${rel_dirname}/${basename} ${rel_dirname}/${basename} root=${linux_root_device_thisversion} $GRUB_CMDLINE_LINUX_DEFAULT EOF if test -n "${initrd}" ; then cat << EOF module ${rel_dirname}/${initrd} ${rel_dirname}/${initrd} EOF fi cat << EOF } EOF list=`echo $list | tr ' ' '\n' | grep -vx $linux | tr '\n' ' '` done -Bruce------------------------------------------------------------------------_______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxxxxxxxx <mailto:Xen-devel@xxxxxxxxxxxxxxxxxxx> <mailto:Xen-devel@xxxxxxxxxxxxxxxxxxx <mailto:Xen-devel@xxxxxxxxxxxxxxxxxxx>> http://lists.xensource.com/xen-devel------------------------------------------------------------------------ _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxxxxxxxx <mailto:Xen-devel@xxxxxxxxxxxxxxxxxxx> http://lists.xensource.com/xen-devel------------------------------------------------------------------------ _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxxxxxxxx http://lists.xensource.com/xen-devel _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxxxxxxxx http://lists.xensource.com/xen-devel
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |