Arch Linux User Repository

I no longer use dracut, and do not maintain this hook anymore.

People here should switch to dracut-ukify which uses systemd's ukify tool and provides similar experience.

I use mkinitcpio and kernel-install. I wouldn't like to make any recommendations; this choice just reflects my personal preferences.

Please note that adopting this PKGBUILD isn't enough; the upstream repo was also mine and is now archived. You may want to fork it or move the sources back here into this repo.

@ewout I think it's booster: https://wiki.archlinux.org/title/Booster

@swsnr What mkinitpcio/UKI setup do you use now btw? Do you recommend migrating to something else?

@swsnr Adopted the package, as I still rely on it. 👍

I no longer use dracut, and do not maintain this hook anymore.

I moved the code to Github at https://github.com/swsnr/dracut-hook-uefi because this is easier for me than AUR comments.

If you've got problems with this package please open an issue at https://github.com/swsnr/dracut-hook-uefi/issues/new instead of writing a comment here.

Thanks

Interesting, indeed. Sounds reasonable, but I wonder why I never had this issue with this setup, even though that's exactly how my systems are set up.

I have no real idea how to fix this, and given that I can't reproduce it, I can't even test it. So I'd appreciate a patch to fix this problem.

Hit an interesting case: system using discoverable partitions, has no /etc/fstab file but has a /efi dir, thus it is a systemd.automount; when upgrading kernels, and very likely I did not look at /efi before, the action to trigger mounting will be performed in PostTransaction, but vfat.ko for the running kernel is already removed. I think it may be fixed by a stat /efi/something in PreTransaction.

As an update, a recent update to dkms seems to have fixed the issue with the image generation so it isn't really a problem anymore.