I suddenly got the same problem in /efi/loader/entries all conf files reverted to previous uuid (first post ). To change uuid I just edited all conf files with the correct uuid for root.
I don't know if manually changing was correct or if I should use some sort command.
I presume these are filesystem UUIDs. I also presume from your other post, that you used a live USB to fix nvidia drivers? Note that nvidia driver installers/packages trigger a initrd rebuild, and if you do that in a live environment, it’s possible that you will get the UUID of your live USB filesystem and not your actual boot drive… at least that’s my guess.
If you booted into a live USB you need to make sure that you chroot into your install on your disk whenever doing any operations on the boot loader. That involves mounting your actual disk (eg, /dev/nvme0p1) somewhere on the live USB (eg, /mnt/example), then bind-mounting the proc, sys, dev, tempfs filesystems under /mnt/example/proc, /mnt/example/sys, etc. You may also need to mount /efi under /mnt/example/efi or boot/efi (wherever you have it in your system). Next, chroot to /mnt/example. You should now have a fully functional install you normally boot into, with the only difference being that the kernel booted off the USB drive. Now you can try reinstalling drivers, rebuilding initrd, reconfiguring the bootloader, etc. Since you’re chrooted, the system should see the proper UUIDs, in theory…
If you want a more comprehensive tutorial on how to do this, look for bootloader fixing tutorials.
I managed to find a problem and fix it. The problem was that /etc/kernel/cmd had the wrong UUID. Thanks for giving advice about initrd and bootloader reconfiguring, might not have found a solution without it.
Your post is incredibly informative and helpful, so this isn't aimed at you at all, but this kind of fix is why Linux is not ready for the everyday average user.
Meh, it could be done in a repair utility, but there's no central power to distribute it and systems can be setup in too many different ways for it to make sense. This is part of the advanced Linux learning curve, not necessary for regular use. Windows can get hosed as well but requires a reinstall because tools like this are not easily available (or you fix windows with Linux).
To be fair, average users would never (or should never) encounter such an issue. The person asking uses Arch (I think?) which is by far not an “average person” distribution.
I used arch-chroot.
I mounted efi to efi dir with mount and used mount -o subvol=@subvolname to sub vol dir.
Just incase i will reinstall nvidia drivers when booted normally.
I will read about initrd.
Thanks for all the info.
I think the key would be figuring out where this extra UUID is coming from. Maybe next time you try this, make a note of all the UUIDs on your system (including the bootable USB) and see which one ends up in the bootloader config.
Knowing what’s happening can help guide your Googling to find out why it’s happening and how to fix it.