On Jul 29, 2024, at 22:23, Peter Krempa <pkrempa(a)redhat.com>
wrote:
On Mon, Jul 22, 2024 at 15:43:22 +0900, Itaru Kitayama wrote:
> Hi,
> With the below virt-install options, I am trying to
import a Realm VM:
> […]
> export LIBVIRT_QEMU=/home/realm/qemu-system-aarch64
> virt-install --machine=virt --arch=aarch64 --name=test8
--disk
path=/var/lib/libvirt/images/jammy.qcow2,format=qcow2,device=disk,bus=virtio,cache=none
--memory=2048 --vcpu=8 --nographic --check all=off --features acpi=off --import
--os-variant=ubuntu22.04 --qemu-commandline="-M
virt,confidential-guest-support=rme0,acpi=off,gic-version=3 -cpu host -object
rme-guest,id=rme0,measurement-algo=sha512" --boot
kernel=Image-cca,initrd=rootfs.cpio
You're passing the kernel image file and initrd path as relative paths.
Make sure that it gets interpreted correctly. I suggest always using
full paths, due to the fact that libvirt itself is running the VM from a
separate daemon which might interpret it differently.
> […]
> However, it fails with the messages:
> WARNING Disk /var/lib/libvirt/images/jammy.qcow2 is
already in use by other guests ['test', 'tmp1', 'jammy'].
> Starting install...
> ERROR internal error: QEMU unexpectedly closed the monitor (vm='test8'):
2024-07-22T06:40:39.290847Z qemu-system-aarch64: could not load kernel
'/home/realm/Image-cca'
Looking at qemu sources this error seems to be reported when it can't
extract the kernel itself from the given file, so check that the kernel
image is correct and an actual kernel image.
Otherwise libvirt simply passes the path to the kernel image to qemu and
doesn't interpret it in any way so from there I'd suggest contacting
the qemu mailing list.
Ok. Actually I did:
Upon checking the libvirtd status (the one Ubuntu 24.04 provides) I see:
/sys/firmware/efi/systab: SMBIOS entry point missing
This is observed even after replacing the Arm CCA support kernel to upstream v6.10 kernel
with some DMI configuration changes in.
Thanks,
Itaru.