shim-signed fails to upgrade when NVMe drive is moved into the VM as Virtio drive

[Wladimir Mutel]

I had to remove a NVMe storage (nvme0n1) from my laptop and temporarily brought it back as a VirtIO (vda) device under a VM. When I received shim-signed upgrade to version 1.51+15.4-0ubuntu9 , I got this :

Setting up shim-signed (1.51+15.4-0ubuntu9) ...
mount: /var/lib/grub/esp: special device /dev/disk/by-id/nvme-eui.001b448b4993ceb4-part1 does not exist.
dpkg: error processing package shim-signed (--configure):
 installed shim-signed package post-installation script subprocess returned error exit status 32
Errors were encountered while processing:
 shim-signed
E: Sub-process /usr/bin/dpkg returned an error code (1)

Looking into /var/lib/dpkg/info/shim-signed.postinst , I see that it calls /usr/lib/grub/grub-multi-install which in turn gets this device name from debconf DB, which it then fails to find among VM devices.

Looking into /var/cache/debconf/config.dat , I see it recorded under Name: grub-efi/install_devices , also its nvme0n1p1 partition recorded under Name: grub-efi/partition_description .
I am not sure whether it is safe to remove or modify these values by text editor, but running grub-install or dpkg-reconfigure grub-efi-amd64 as root does not change them .

Please help me to resolve this situation.

[Jason C. McDonald (codemouse92)]

I don't believe this is a duplicate. shim-signed was looking for the NVMe storage that was removed and inaccessible. Bug #1940723 has a similar mechanism of failure, but is related to cloud instances, rather than a missing device.

[Steve Langasek (vorlon)]

In terms of shim-signed and grub, the bug is the same. If you think this is a separate bug, then please assign it away from shim-signed to where you think the bug lies, because from a shim-signed perspective it is the same bug.