Failed to find an appropriate bootloader for this system

kro

hello, after an update I had issues with my nvidia graphic card (I know :/) and abandoned drivers with black screen (and still SSH access).
I tried to install other drivers, then nouveau, and long story short I ended breaking my system with no space left on the EFI device :

kro@tour-noire ~ $ sudo eopkg rm nvidia-470-glx-driver-current
The following list of packages will be removed
in the respective order to satisfy dependencies:
nvidia-470-glx-driver-current 
Removing package nvidia-470-glx-driver-current
Removed nvidia-470-glx-driver-current
 [✓] Syncing filesystems                                                success
 [✗] Updating clr-boot-manager                                           failed

A copy of the command output follows:

[FATAL] cbm (../src/bootman/kernel.c:L732): Failed to install initrd /boot/EFI/com.solus-project/initrd-com.solus-project.current.6.14.11-320: No space left on device
[FATAL] cbm (../src/bootman/update.c:L250): Failed to install default-current kernel: ///usr/lib/kernel/com.solus-project.current.6.14.11-320

Then I can't boot anymore (sorry I don't remember the exact message and don't want to reboot right now but it's related to EFI).
I booted a live USB system, enlarge the EFI partition from 512M to 1G, chrooted following this doc.

So right now I'm supposed to be chrooted but trying to install on remove nouveau or ncidia proprietary drivers I still can't modify the ctrl-boot-manager :
[ ] Updating clr-boot-manager skipped (when trying to install or remove nvidia related stuff)

root@solus / # clr-boot-manager update
[FATAL] cbm (../src/bootman/bootman.c:L153): Failed to find an appropriate bootloader for this system

on the host the sda1 / EFI part is (correctly ?) mounted on /target/boot :

live@solus /target/boot/EFI $ lsblk -o NAME,MODEL,SIZE,PATH,MOUNTPOINT
NAME   MODEL                      SIZE PATH       MOUNTPOINT
loop0                             2.8G /dev/loop0 
loop1                             9.6G /dev/loop1 /run/rootfsbase
sda    INTEL SSDSA2M080G2GC      74.5G /dev/sda   
├─sda1                           1002M /dev/sda1  /target/boot
├─sda2                            3.7G /dev/sda2  
└─sda3                           69.8G /dev/sda3  /target
sdb    WDC WD5000BUCT-63LS5Y1   465.8G /dev/sdb   
└─sdb1                          465.8G /dev/sdb1  
sdc    Voyager GT 3.0            58.9G /dev/sdc   
├─sdc1                            3.1G /dev/sdc1  /run/initramfs/live
└─sdc2                            200M /dev/sdc2  
sr0    HL-DT-ST DVDRAM GH22NS70  1024M /dev/sr0   
zram0                             3.8G /dev/zram0 [SWAP]

What gives ?

How could I run clr-boot-manager update from this USB live system ?

alfisya

Have you tried following the boot rescue guide? If you did here are few discussion threads that might be of your interest:

There are more if you search on forum with "No boot" keyword. Hopefully you get it solved, cheers!

kro

yes I linked it in my previous message.

actually I already have done a few searches on the forum, but haven't found anything related to my specific issue I believe...

root@solus / # clr-boot-manager update
[FATAL] cbm (../src/bootman/bootman.c:L153): Failed to find an appropriate bootloader for this system

kro

I might have to recreate the EFI stuff in its partition I guess.

I found this https://discuss.getsol.us/d/3681-hdd-efi-bootloader-mia-post-usb-install/4

sudo efibootmgr -c -d /dev/sda -p 1 -l "\EFI\systemd\systemd-bootx64.efi" -L "Linux Boot Manager"

Do I need to run this from the chrooted environment or from anywhere ?
(efibootmgr command only seems to exist outside the chrooted env however ... binary sits in the /usr space which might not be linked I guess)

sorry it seems I can't find any "official" documentation on how to rebuild the efi bootloader https://www.google.com/search?q=site%3Agetsol.us+rebuild+efi+bootloader+efibootmgr

brent

kro Do I need to run this from the chrooted environment or from anywhere ?

yes.
however for borked bootloaders (i've had some) as long as I was LIVE and chrooted into /boot /root the boot would rebuild/repopulate itself without that specific command you have there (using the commands in the tutorial instead).

I think you are on the right path. but that guy in the thread had a problem that had nothing to do with NVIDIA that I could see.
2 cents

kro

I installed efibootmgr from the chooted terminal then.

but

root@solus ~ # efibootmgr -c -d /dev/sda -p 1 -l "\EFI\systemd\systemd-bootx64.efi" -L "Linux Boot Manager"
EFI variables are not supported on this system.

OK I think the problem with this live USB session must be coming from the fact I must have booted from the BIOS (legacy ?) and not UEFI !

kro

so that was an issue indeed.
I'm on UEFI now !

but then :

root@solus / # clr-boot-manager update
[ERROR] cbm (../src/bootman/update.c:L189): Cannot determine the currently running kernel

though :

root@solus / # clr-boot-manager list-kernels
* com.solus-project.current.6.14.11-320
  com.solus-project.current.6.11.7-308
  com.solus-project.lts.6.12.34-270
  com.solus-project.lts.6.6.60-259
  com.solus-project.current.6.4.15-254
  com.solus-project.lts.5.15.131-215

... weird !

following this message I thought I had a missing /usr/lib/kernel/current-kernel
so I created it :
ln -s /usr/lib/kernel/current-vmlinux /usr/lib/kernel/current-kernel
and still have the same error message for clr-boot-manager update.

another weird thing :

root@solus /usr/lib/kernel # uname -r
6.12.9-312.current

I guess this difference between supposed active kernel(s) must be due to the chroot ?
(both commands are run from the chrooted terminal though)

kro

after rebooting (or maybe because of the symlink I've added, I'm not sure),
the message slightly changes :
root@solus / # clr-boot-manager update [FATAL] cbm (../src/bootman/bootman.c:L548): Cannot determine boot device

// edit : no I just forgot to mount the EFI partition this time !
current error is still Cannot determine the currently running kernel.

kro

I fixed the boot !

root@solus / # efibootmgr -c -d /dev/sda -p 1 -l "\EFI\systemd\systemd-bootx64.efi" -L "Linux Boot Manager"
efibootmgr: ** Warning ** : Boot0000 has same label Linux Boot Manager
BootCurrent: 0004
Timeout: 1 seconds
BootOrder: 0006,0005,0003,0004,0001,0002,0000
Boot0000  Linux Boot Manager	HD(1,GPT,76eadfc2-43c4-4317-b7db-4d28d1188964,0x40,0xf4240)/\EFI\systemd\systemd-bootx64.efi
Boot0001  Hard Drive	BBS(HD,,0x0)
Boot0002  CD/DVD Drive	BBS(CDROM,,0x0)50333a20484c2d44542d53542044564452414d20474832324e533730202000
Boot0003* Solus Linux Bootloader	HD(1,GPT,76eadfc2-43c4-4317-b7db-4d28d1188964,0x40,0xf4240)/\EFI\com.solus-project\bootloaderx64.efi
Boot0004* UEFI: Corsair Voyager GT 3.0 000B	PciRoot(0x0)/Pci(0x1c,0x4)/Pci(0x0,0x0)/USB(2,0)/HD(1,MBR,0x5bedc865,0x88,0x64000)
Boot0005* Solus Linux Bootloader	HD(1,GPT,76eadfc2-43c4-4317-b7db-4d28d1188964,0x40,0x1f4fc0)/\EFI\com.solus-project\bootloaderx64.efi
Boot0006* Linux Boot Manager	HD(1,GPT,76eadfc2-43c4-4317-b7db-4d28d1188964,0x40,0x1f4fc0)/\EFI\systemd\systemd-bootx64.efi
root@solus / # clr-boot-manager update
[ERROR] cbm (../src/bootman/update.c:L189): Cannot determine the currently running kernel

The issue was indeed due a broken UEFI partition I was not able to fix while booting a non UEFI live session.

now I still have issues with my GPU but I can SSH in, so that will be for another topic, if any.

Thank you for your feedbacks and support, and for the fixing boot guide to whoever edits it, it has been a great help !

brent

kro you learned a valuable lesson about Legacy and UEFI incompatibility. It's bit us all in the a** at one point. Congrats for sticking with your troubleshooting and being up and running again.