Multipass, could not load PC BIOS 'OVMF.fd'

Support AUR
1

Hi

I have an issue getting Multipass working? It’s an AUR package.
When I try to run it, I get

$ sudo multipass launch 22.04 -vvvv
.
.
.
[2024-11-03T10:52:44.802] [debug] [informed-wigeon] process working dir ''
[2024-11-03T10:52:44.802] [info] [informed-wigeon] process program 'qemu-system-x86_64'
[2024-11-03T10:52:44.802] [info] [informed-wigeon] process arguments '-bios, OVMF.fd, --enable-kvm, -cpu, host, -nic, tap,ifname=tap-1ec20fe51e3,script=no,downscript=no,model=virtio-net-pci,mac=52:54:00:a8:ad:48, -device, virtio-scsi-pci,id=scsi0, -drive, file=/var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/ubuntu-22.04-server-cloudimg-amd64.img,if=none,format=qcow2,discard=unmap,id=hda, -device, scsi-hd,drive=hda,bus=scsi0.0, -smp, 1, -m, 1024M, -qmp, stdio, -chardev, null,id=char0, -serial, chardev:char0, -nographic, -L, /usr/share/ovmf/, -L, /usr/share/ovmf/x64/, -L, /usr/share/qemu/, -cdrom, /var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/cloud-init-config.iso'
[2024-11-03T10:52:44.810] [debug] [qemu-system-x86_64] [670584] started: qemu-system-x86_64 -nographic -dump-vmstate /tmp/multipassd.ukSPNS
[2024-11-03T10:52:44.851] [debug] [daemon] Applied AppArmor policy: multipass.informed-wigeon.qemu-system-x86_64
[2024-11-03T10:52:44.852] [info] [informed-wigeon] process state changed to Starting
[2024-11-03T10:52:44.860] [info] [informed-wigeon] process state changed to Running
[2024-11-03T10:52:44.861] [debug] [informed-wigeon] Adding firmware directory: /usr/share/ovmf/
[2024-11-03T10:52:44.861] [debug] [informed-wigeon] Adding firmware directory: /usr/share/ovmf/x64/
[2024-11-03T10:52:44.861] [debug] [informed-wigeon] Adding firmware directory: /usr/share/qemu/
[2024-11-03T10:52:44.861] [debug] [qemu-system-x86_64] [670589] started: qemu-system-x86_64 -bios OVMF.fd --enable-kvm -cpu host -nic tap,ifname=tap-1ec20fe51e3,script=no,downscript=no,model=virtio-net-pci,mac=52:54:00:a8:ad:48 -device virtio-scsi-pci,id=scsi0 -drive file=/var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/ubuntu-22.04-server-cloudimg-amd64.img,if=none,format=qcow2,discard=unmap,id=hda -device scsi-hd,drive=hda,bus=scsi0.0 -smp 1 -m 1024M -qmp stdio -chardev null,id=char0 -serial chardev:char0 -nographic -L /usr/share/ovmf/ -L /usr/share/ovmf/x64/ -L /usr/share/qemu/ -cdrom /var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/cloud-init-config.iso
[2024-11-03T10:52:44.862] [info] [informed-wigeon] process started
launch failed: The following errors occurred:
qemu: could not load PC BIOS 'OVMF.fd'
informed-wigeon: shutdown called while starting

The file is present in the search folder:

$ ls -lr /usr/share/ovmf/x64/
total 29408
-rw-r--r-- 1 root root  131072 nov.  24  2023 OVMF_VARS.fd
-rw-r--r-- 1 root root  540672 nov.  24  2023 OVMF_VARS.4m.fd
-rw-r--r-- 1 root root 2097152 nov.  24  2023 OVMF.fd
-rw-r--r-- 1 root root 1966080 nov.  24  2023 OVMF_CODE.secboot.fd
-rw-r--r-- 1 root root 3653632 nov.  24  2023 OVMF_CODE.secboot.4m.fd
-rw-r--r-- 1 root root 1966080 nov.  24  2023 OVMF_CODE.fd
-rw-r--r-- 1 root root 1966080 nov.  24  2023 OVMF_CODE.csm.fd
-rw-r--r-- 1 root root 3653632 nov.  24  2023 OVMF_CODE.csm.4m.fd
-rw-r--r-- 1 root root 3653632 nov.  24  2023 OVMF_CODE.4m.fd
-rw-r--r-- 1 root root 4194304 nov.  24  2023 OVMF.4m.fd
-rw-r--r-- 1 root root 2097152 nov.  24  2023 MICROVM.fd
-rw-r--r-- 1 root root 4194304 nov.  24  2023 MICROVM.4m.fd

The same issue is described here
The aur package has a patch in place resolving this, AUR
On my ArchLinux installation, Multipass works. But on Manjaro it doesn’t. Looking on the log, it seems like the patch is applied and multipass is searching the added paths for the file. And a ls even shows that the file is present.

Initially multipass only checks /usr/share/qemu/ for the bios file. But the patch in the AUR package adds /usr/share/ovmf/x64/ to the search path. In the github issue, linking the files from the ovmf folder til the qemu folder was suggested as a working fix, but it didn’t work for me. I’ve even tried to copy the bios files to to the qemu folder. But no luck.

Have anyone got Multipass working? Or does anyone have any tips on what to try next?

Orjanp

2

Switch to unstable branch.

Manjaro Unstable == ArchLinux Stable

3

You just listed the bug and the hack to fix in right in your own post.

Your firmware files:

pacman -Ql edk2-ovmf | grep OVMF_CODE.fd                                                            
edk2-ovmf /usr/share/edk2/ia32/OVMF_CODE.fd
edk2-ovmf /usr/share/edk2/x64/OVMF_CODE.fd

You passed it: -L /usr/share/ovmf/x64/

Fix the parameter you pass, if you can’t, they were just doing something like:

sudo ln -s /usr/share/edk2/ /usr/share/ovmf
4

Thanks for the suggestions.

It turns out /usr/share/ovmf is actually a link to /usr/share/edk2

lrwxrwxrwx   1 root     root        15 nov.  24  2023 ovmf -> /usr/share/edk2
lrwxrwxrwx   1 root     root        15 nov.  24  2023 OVMF -> /usr/share/edk2

I tried running the command that is failing manually.

$ sudo qemu-system-x86_64 -bios OVMF.fd --enable-kvm -cpu host -nic tap,ifname=tap-1ec20fe51e3,script=no,downscript=no,model=virtio-net-pci,mac=52:54:00:a8:ad:48 -device virtio-scsi-pci,id=scsi0 -drive file=/var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/ubuntu-22.04-server-cloudimg-amd64.img,if=none,format=qcow2,discard=unmap,id=hda -device scsi-hd,drive=hda,bus=scsi0.0 -smp 1 -m 1024M -qmp stdio -chardev null,id=char0 -serial chardev:char0 -nographic -L /usr/share/ovmf/ -L /usr/share/ovmf/x64/ -L /usr/share/qemu/ -cdrom /var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/cloud-init-config.iso
{"QMP": {"version": {"qemu": {"micro": 0, "minor": 1, "major": 9}, "package": ""}, "capabilities": ["oob"]}}

And it worked. Removing all the -L options makes it crash.

$ sudo qemu-system-x86_64 -bios OVMF.fd --enable-kvm -cpu host -nic tap,ifname=tap-1ec20fe51e3,script=no,downscript=no,model=virtio-net-pci,mac=52:54:00:a8:ad:48 -device virtio-scsi-pci,id=scsi0 -drive file=/var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/ubuntu-22.04-server-cloudimg-amd64.img,if=none,format=qcow2,discard=unmap,id=hda -device scsi-hd,drive=hda,bus=scsi0.0 -smp 1 -m 1024M -qmp stdio -chardev null,id=char0 -serial chardev:char0 -nographic -cdrom /var/lib/multipassd/.local/share/multipassd/vault/instances/informed-wigeon/cloud-init-config.iso
{"QMP": {"version": {"qemu": {"micro": 0, "minor": 1, "major": 9}, "package": ""}, "capabilities": ["oob"]}}
qemu: could not load PC BIOS 'OVMF.fd'

Since Multipass is using the command with the -L option, I guess it should work.

I checked out the difference of the version of the dependencies between the systems.

# Manjaro
$ sudo pacman -Q apparmor qt5-base qemu-base
apparmor 3.1.7-4
qt5-base 5.15.15+kde+r127-1
qemu-base 9.1.0-2

# ArchLinux
$ sudo pacman -Q apparmor qt5-base qemu-base
apparmor 4.0.3-1
qt5-base 5.15.15+kde+r136-1
qemu-base 9.1.1-1

My next step will be to upgrade Apparmor to version 4 to see if that makes a difference.

5

Yes, that might be the solution in the end. I didn’t realize it was that easy switching back and forth. Not sure how upgrading only the individual Multipass dependencies to a newer version will affect the whole system so maybe a full switch to unstable is a better option.

6

Switched to unstable, updated, rebooted and rebuilt multipass, but no luck. Still the same error. Not sure where to go from here.