GRUB/Refind creating no persistent NVRAM entries.

[daguq]

I am running the command

Code: Select all

grub-install --target=x86_64-efi --efi-directory=/boot --bootloader-id=gentoo --recheck

, and checking that an NVRAM entry is indeed created by running

Code: Select all

efibootmgr -v

and getting the output

Code: Select all

 
Boot0000* gentoo	HD(1,GPT,8dcf8c78-86c1-224f-86ab-43cd95d530fa,0x800,0x100000)/File(\EFI\gentoo\grubx64.efi)
      dp: 04 01 2a 00 01 00 00 00 00 08 00 00 00 00 00 00 00 00 10 00 00 00 00 00 78 8c cf 8d c1 86 4f 22 86 ab 43 cd 95 d5 30 fa 02 02 / 04 04 34 00 5c 00 45 00 46 00 49 00 5c 00 67 00 65 00 6e 00 74 00 6f 00 6f 00 5c 00 67 00 72 00 75 00 62 00 78 00 36 00 34 00 2e 00 65 00 66 00 69 00 00 00 / 7f ff 04 00

Similar issue persists with refind-install

However, when I reboot and enter UEFI interface, this entry is not present there. How can I fix this error?

Steps I have tried so far

1. Mount efivars as rw with

Code: Select all

mount -o remount,rw -t efivarfs efivarfs /sys/firmware/efi/efivars

and run grub-install/refind-install
2. Reset UEFI
3. Update UEFI

[alamahant]

However, when I reboot and enter UEFI interface, this entry is not present there. How can I fix this error?

Just reboot into your Gentoo not the bios/uefi and check again with efibootmgr.

[daguq]

However, when I reboot and enter UEFI interface, this entry is not present there. How can I fix this error?

Just reboot into your Gentoo not the bios/uefi and check again with efibootmgr.

I have done so, and it shows this

Code: Select all

BootCurrent: 0001
Timeout: 0 seconds
BootOrder: 0001
Boot0001* debian	HD(1,GPT,ab95c874-da9a-41d0-bdd0-8631639d2857,0x800,0x100000)/File(\EFI\debian\shimx64.efi) File(.)
      dp: 04 01 2a 00 01 00 00 00 00 08 00 00 00 00 00 00 00 00 10 00 00 00 00 00 74 c8 95 ab 9a da d0 41 bd d0 86 31 63 9d 28 57 02 02 / 04 04 34 00 5c 00 45 00 46 00 49 00 5c 00 64 00 65 00 62 00 69 00 61 00 6e 00 5c 00 73 00 68 00 69 00 6d 00 78 00 36 00 34 00 2e 00 65 00 66 00 69 00 00 00 / 7f ff 04 00
    data: 00 00 42 4f

I have debian installed in a separate disk, with a separate esp.

[alamahant]

Code: Select all

I have debian installed in a separate disk, with a separate esp.

Is Debian's your main initial disk?
Ok it would be best to use the debian ESP for Gentoo also.
Modify Gentoo fstab to mount debian esp rerun grub-install and re-update grub and initramfs.
Then you can "decommission" Gentoo ESP.
Also maybe best NOT to mix grub and refind.

[daguq]

Code: Select all

I have debian installed in a separate disk, with a separate esp.

Is Debian's your main initial disk?
Ok it would be best to use the debian ESP for Gentoo also.
Modify Gentoo fstab to mount debian esp rerun grub-install and re-update grub and initramfs.

I will try and do that.

Any guesses on why the NVRAM entry is not persistent across reboots? This issue seems to only occur with Gentoo, not with other linux distros.

[alamahant]

Any guesses on why the NVRAM entry is not persistent across reboots? This issue seems to only occur with Gentoo, not with other linux distros.

Is Gentoo;s disk an external one?
Very often this happens with external disks and esps ie efibootmgr entries do not persist.

[daguq]

Any guesses on why the NVRAM entry is not persistent across reboots? This issue seems to only occur with Gentoo, not with other linux distros.

Is Gentoo;s disk an external one?
Very often this happens with external disks and esps ie efibootmgr entries do not persist.

No, it is not an external disk.

[alamahant]

I see you mount esp on /boot.
Debian mounts it on /boot/efi
I suggest you create a /boot/efi directory in Gentoo also and mount the debian esp on /boot/efi.
Plz run

Code: Select all

umount /boot
mkdir /boot/efi
mount <debian-esp> /boot/efi
grub-install --target=x86_64-efi --efi-directory=/boot/efi --bootloader-id=gentoo --recheck

You will need to reinstall the kernel because the kernel resided in Gentoo esp which you are no longer using.

and dont forget to update your fstab and grub and initramfs .

Code: Select all

nano /etc/fstab............
cd /usr/src/linux
make install
#update/recreate initramfs
grub-mkconfig -o /boot/grub/grub.cfg

This is why I personally hate mounting ESP on /boot.
Because this way the kernel and initramfs reside inside the ESP.
To me this is wrong.
By the way you can mount esp anywhere you wish.
Maybe @ /efi also possible.

[daguq]

I have managed to get it work by using debian's bootloader.

But it still doesn't resolve my issue. I don't understand why grub/refind isn't creating an nvram entry when run from gentoo, and provided gentoo has it's own esp.

I just installed Fedora and Ubuntu on different drives, with different esp's and both of those NVRAM entries show up in the UEFI boot menu.

[alamahant]

Is maybe secureboot or similar enabled in uefi/bios.
I see debian is using shimx64.efi.Fedora probably does the same.
Gentoo is using grubx64.efi.
Just a foolish thought.

[daguq]

Is maybe secureboot or similar enabled in uefi/bios.
I see debian is using shimx64.efi.Fedora probably does the same.
Gentoo is using grubx64.efi.
Just a foolish thought.

Secureboot is disabled

[sMueggli]

Some UEFI implementations are cleaning "invalid" boot entries to save space. The way how the firmware is doing the cleaning varies from implementation to implementation. It is possible that the firmware does not recognise the Gentoo boot entry as "valid".

Please show the entire output of

Code: Select all

efibootmgr
lsblk -o NAME,PARTTYPE,FSTYPE,PARTUUID

Some personal recommendations: Use only one ESP for the whole hardware platform and mount it to /boot/efi (your firmware does not need access to your kernel and initramfs; do not forget to move the files from /boot to the new /boot).

[daguq]

Some UEFI implementations are cleaning "invalid" boot entries to save space. The way how the firmware is doing the cleaning varies from implementation to implementation. It is possible that the firmware does not recognise the Gentoo boot entry as "valid".

Please show the entire output of

Code: Select all

efibootmgr
lsblk -o NAME,PARTTYPE,FSTYPE,PARTUUID

Some personal recommendations: Use only one ESP for the whole hardware platform and mount it to /boot/efi (your firmware does not need access to your kernel and initramfs; do not forget to move the files from /boot to the new /boot).

Entire output of

Code: Select all

efibootmgr

after running

Code: Select all

grub-install

Code: Select all

BootCurrent: 0000
Timeout: 0 seconds
BootOrder: 0001,0000,0003,0004,0005,0006,0007
Boot0000* ubuntu        HD(1,GPT,6f0d2fc6-d8c2-9d42-88b1-4f369f37248b,0x800,0x219800)/File(\EFI\ubuntu\shimx64.efi)
Boot0001* gentoo        HD(1,GPT,fd65a76f-3e5a-334f-b0c2-13b7d7523126,0x800,0x200000)/File(\EFI\gentoo\grubx64.efi)
Boot0003* UEFI:  USB, Partition 2       PciRoot(0x0)/Pci(0x14,0x0)/USB(24,0)/HD(2,MBR,0x9bd7848e,0x7289000,0x10000)0000424f
Boot0004* UEFI:  USB, Partition 1       PciRoot(0x0)/Pci(0x14,0x0)/USB(24,0)/HD(1,MBR,0x9bd7848e,0x800,0x7288800)0000424f
Boot0005* UEFI:CD/DVD Drive     BBS(129,,0x0)
Boot0006* UEFI:Removable Device BBS(130,,0x0)
Boot0007* UEFI:Network Device   BBS(131,,0x0)

After reboot

Code: Select all

BootCurrent: 0000
Timeout: 0 seconds
BootOrder: 0000
Boot0000* ubuntu        HD(1,GPT,6f0d2fc6-d8c2-9d42-88b1-4f369f37248b,0x800,0x219800)/File(\EFI\ubuntu\shimx64.efi)

Output of

Code: Select all

 lsblk -o NAME,PARTTYPE,FSTYPE,PARTUUID

Code: Select all

NAME        PARTTYPE                             FSTYPE PARTUUID
nvme0n1
├─nvme0n1p1 c12a7328-f81f-11d2-ba4b-00a0c93ec93b vfat   fd65a76f-3e5a-334f-b0c2-13b7d7523126
└─nvme0n1p2 0fc63daf-8483-4772-8e79-3d69d8477de4 ext4   8ed23a45-a752-0540-b824-efd93ce9c9e5
nvme1n1
├─nvme1n1p1 c12a7328-f81f-11d2-ba4b-00a0c93ec93b vfat   6f0d2fc6-d8c2-9d42-88b1-4f369f37248b
└─nvme1n1p2 0fc63daf-8483-4772-8e79-3d69d8477de4 ext4   3712929c-b78e-3343-acb4-1188957cef6b

[alamahant]

As both I and sMueggli pointed out plz consider using only one ESP.
Or use os-prober as you have to boot the other OSes.
Ubuntu now?
What happened to debian?

[daguq]

As both I and sMueggli pointed out plz consider using only one ESP.
Or use os-prober as you have to boot the other OSes.
Ubuntu now?
What happened to debian?

I nuked both debian and gentoo, to install fedora and ubuntu to check if both of them can boot with separate esp's. It turns out they can, and they have seperate NVRAM entries in UEFI too.

Moving on, I nuked the fedora install to reinstall gentoo, and as it turns out, nothing registers the NVRAM entry for gentoo given it's own esp.

I could use a single ESP for both gentoo and ubuntu and call it a day, but that's avoiding the problem. I am interested in why no NVRAM entry is being created for gentoo with it's own esp, and it's only happening with Gentoo, not with other distros.

[pietinger]

daguq,

using only one ESP (efi system partition) is a good advice from @alamahant and @sMueggli. If you want two ESP then check if BOTH have the ESP-flag enabled.

e.g. with parted (example from my machine):

Code: Select all

# parted -l
Modell: Samsung SSD 980 PRO 1TB (nvme)
Festplatte  /dev/nvme0n1:  1000GB
Sektorgröße (logisch/physisch): 512B/512B
Partitionstabelle: gpt
Disk-Flags: 

Nummer  Anfang  Ende    Größe   Dateisystem     Name  Flags
 1      1049kB  268MB   267MB   fat32           Boot  boot, esp
 2      268MB   4295MB  4027MB  linux-swap(v1)  Swap
 3      4295MB  1000GB  996GB   ext4            Root

If you miss the ESP-flag on your 2nd ESP your UEFI will never look into your 2nd ESP. You can enable it with:

Code: Select all

# parted /dev/nvme0n1 set 1 boot on
OR
# parted /dev/nvme1n1 set 1 boot on

(After this you must create your missing UEFI entry for gentoo's grub again with "efibootmgr"; see more here:
https://wiki.gentoo.org/wiki/Efibootmgr ... boot_entry )

[daguq]

daguq,

using only one ESP (efi system partition) is a good advice from @alamahant and @sMueggli. If you want two ESP then check if BOTH have the ESP-flag enabled.

e.g. with parted (example from my machine):

Code: Select all

# parted -l
Modell: Samsung SSD 980 PRO 1TB (nvme)
Festplatte  /dev/nvme0n1:  1000GB
Sektorgröße (logisch/physisch): 512B/512B
Partitionstabelle: gpt
Disk-Flags: 

Nummer  Anfang  Ende    Größe   Dateisystem     Name  Flags
 1      1049kB  268MB   267MB   fat32           Boot  boot, esp
 2      268MB   4295MB  4027MB  linux-swap(v1)  Swap
 3      4295MB  1000GB  996GB   ext4            Root

If you miss the ESP-flag on your 2nd ESP your UEFI will never look into your 2nd ESP. You can enable it with:

Code: Select all

# parted /dev/nvme0n1 set 1 boot on
OR
# parted /dev/nvme1n1 set 1 boot on

(After this you must create your missing UEFI entry for gentoo's grub again with "efibootmgr"; see more here:
https://wiki.gentoo.org/wiki/Efibootmgr ... boot_entry )

Here's the output of

Code: Select all

 parted -l

Code: Select all

Model: WD Blue SN570 1TB (nvme)
Disk /dev/nvme0n1: 1000GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags: 

Number  Start   End     Size    File system  Name  Flags
 1      1049kB  1075MB  1074MB  fat32              boot, esp
 2      1075MB  1000GB  999GB   ext4


Model: WD Blue SN570 1TB (nvme)
Disk /dev/nvme1n1: 1000GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags: 

Number  Start   End     Size    File system  Name  Flags
 1      1049kB  1128MB  1127MB  fat32              boot, esp
 2      1128MB  1000GB  999GB   ext4

I agree that having a single esp is a good idea, but I can't understand why no NVRAM entry is being created in the UEFI if I give gentoo it's own ESP. It's weird behavior, don't you agree?

I will try once more with efibootmgr, and report.

[daguq]

No luck with

Code: Select all

efibootmgr

I ran

Code: Select all

efibootmgr -c -d /dev/nvme0n1 -p 2 -L "Gentoo" -l "\efi\boot\bootx64.efi" initrd='\initramfs-6.1.41-gentoo-dist'

and the output was

Code: Select all

BootCurrent: 0000
Timeout: 0 seconds
BootOrder: 0001,0000
Boot0000* ubuntu	HD(1,GPT,6f0d2fc6-d8c2-9d42-88b1-4f369f37248b,0x800,0x219800)/File(\EFI\ubuntu\shimx64.efi)
Boot0001* Gentoo	HD(2,GPT,8ed23a45-a752-0540-b824-efd93ce9c9e5,0x200800,0x74506000)/File(\efi\boot\bootx64.efi)initrd=\initramfs-6.1.41-gentoo-dist

I rebooted, but no NVRAM entry in UEFI.

[pietinger]

I agree that having a single esp is a good idea, but I can't understand why no NVRAM entry is being created in the UEFI if I give gentoo it's own ESP. It's weird behavior, don't you agree?

I see you have only two partitions for gentoo: One ESP and a big /root. This means: Your /boot-directory is NOT an extra partitions (it is part of your root file-system/-partition). So, how have you mounted your ESP ?
If you use /boot/efi/EFI/.... you must install grub with:

Code: Select all

grub-install --target=x86_64-efi --efi-directory=/boot/efi

to get it in the correct directory.

I assume that you dont have your grubx64.efi in your ESP ... and then your BIOS just delete the entry because it is invalid (== there is no \EFI\gentoo\grubx64.efi). Just check the content of your ESP.

(P.S.: With efibootmgr version 18 you dont need -v anymore; just do a "efibootmgr")

[pietinger]

I ran

Code: Select all

efibootmgr -c -d /dev/nvme0n1 -p 2 -L "Gentoo" -l "\efi\boot\bootx64.efi" initrd='\initramfs-6.1.41-gentoo-dist'

This willl work only IF you have in your ESP the file: /efi/boot/bootx64.efi
Usually Grub is in /efi/gentoo/grubx64.efi

[daguq]

I agree that having a single esp is a good idea, but I can't understand why no NVRAM entry is being created in the UEFI if I give gentoo it's own ESP. It's weird behavior, don't you agree?

I see you have only two partitions for gentoo: One ESP and a big /root. This means: Your /boot-directory is NOT an extra partitions (it is part of your root file-system/-partition). So, how have you mounted your ESP ?
If you use /boot/efi/EFI/.... you must install grub with:

Code: Select all

grub-install --target=x86_64-efi --efi-directory=/boot/efi

to get it in the correct directory.

I assume that you dont have your grubx64.efi in your ESP ... and then your BIOS just delete the entry because it is invalid (== there is no \EFI\gentoo\grubx64.efi). Just check the content of your ESP.

(P.S.: With efibootmgr version 18 you dont need -v anymore; just do a "efibootmgr")

my ESP is mounted at /boot/efi and grubx64.efi is indeed present in \EFI\gentoo\

[daguq]

I ran

Code: Select all

efibootmgr -c -d /dev/nvme0n1 -p 2 -L "Gentoo" -l "\efi\boot\bootx64.efi" initrd='\initramfs-6.1.41-gentoo-dist'

This willl work only IF you have in your ESP the file: /efi/boot/bootx64.efi
Usually Grub is in /efi/gentoo/grubx64.efi

I created the file according to the instructions in the installation handbook

[pietinger]

my ESP is mounted at /boot/efi and grubx64.efi is indeed present in \EFI\gentoo\

Then try to create the correct entry with:

Code: Select all

efibootmgr -c -d /dev/nvme0n1 -p 1 -L "Gentoo" -l "\efi\gentoo\grubx64.efi"

[daguq]

my ESP is mounted at /boot/efi and grubx64.efi is indeed present in \EFI\gentoo\

Then try to create the correct entry with:

Code: Select all

efibootmgr -c -d /dev/nvme0n1 -p 1 -L "Gentoo" -l "\efi\gentoo\grubx64.efi"

I did this and the output is

Code: Select all

BootCurrent: 0000
Timeout: 0 seconds
BootOrder: 0001,0000
Boot0000* ubuntu	HD(1,GPT,6f0d2fc6-d8c2-9d42-88b1-4f369f37248b,0x800,0x219800)/File(\EFI\ubuntu\shimx64.efi)
Boot0001* Gentoo	HD(1,GPT,fd65a76f-3e5a-334f-b0c2-13b7d7523126,0x800,0x200000)/File(\efi\gentoo\grubx64.efi)

I rebooted, but still no NVRAM entry.

[pietinger]

Okay, let us check all requirements:

1. SecureBoot is OFF (yes)
2. Both NVMEs have GPT (yes)
3. All ESP have ESP-flag enabled (yes)
4. All ESP are formatted with FAT32 (yes)
5. In your first ESP (of your first NVMe) you have \efi\gentoo\grubx64.efi (yes)
6. UEFI entry is pointing to it (yes)
7. Your machine is able to use two ESP (you said you have checked this with two ohter Linux distros)

... There is nothing more ...

... What happens if you REMOVE your 2nd NVME completely from your mainboard - only for testing ? (maybe create again an entry for Gentoo BEFORE rebooting; maybe from booting with a LiveCD)