Package: grub-common
Version: 2.04-5
When update-grub is run from the Debian bullseye/sid system, 30_os-prober
detects another OS (in this case Arch Linux) on a second drive. However,
that Arch Linux system uses early loading of microcode updates - see:
https://wiki.archlinux.org/index.php/Microcode
The grub.cfg generated on the Arch Linux disk provides two arguments to
initrd:
initrd /boot/intel-ucode.img /boot/initramfs-linux.img
The grub.cfg generated by the Debian system produces an initrd for the Arch
Linux boot selection that contains only:
initrd /boot/intel-ucode.img
As one might expect, attempting to boot Arch Linux from the grub boot
menu selection generated by the Debian system results in:
Kernel panic - not syncing: VFS: unable to mount root fs on unknown
block(0,0)
The fix would seem to be to make sure update-grub (or whatever it calls)
reads
all arguments to initrd.
Manually editing grub.cfg on the Debian system adding the second argument to
initrd solves the problem and the second OS boots as expected.
Acknowledgement sent
to Krzmbrzl <[email protected]>:
Extra info received and forwarded to list. Copy sent to Debian Install System Team <[email protected]>.
(Mon, 01 May 2023 12:45:03 GMT) (full text, mbox, link).
Subject: update-grub fails to process more than one argument to initrd
Date: Mon, 1 May 2023 14:42:41 +0200
Hi everyone,
In this report, there seems to be an indication that this issue was
fixed with grub2/2.04-5. At least that's how I interpret the
auto-generated line "No longer marked as found in versions grub2/2.04-5"
that was somehow triggered by Colin Watson (though when checking the
respective mail, it seems like that is only about reassignment and not
about a resolution).
In either case, I want to report that this issue still exists when
having grub2-common/2.06 installed (on Ubuntu 22.04), which comes with
os-prober/1.79.
Furthermore, I can explain exactly where this issue occurs. I have
written up a question on the Unix&Linux StackExchange about this, that
contains the explanation: https://unix.stackexchange.com/q/744624/203826
The gist of it is that the script that parses the grub.cfg on the
separate Arch (in my case Manjaro) partition, the initrd statement is
parsed as expected, but the code only handles the first argument to the
initrd command, which in Arch's case is the microcode stuff, while
completely ignoring any subsequent arguments.
Where can I hand in a patch to fix this issue?
Acknowledgement sent
to Steve McIntyre <[email protected]>:
Extra info received and forwarded to list. Copy sent to Debian Install System Team <[email protected]>.
(Tue, 02 May 2023 11:30:02 GMT) (full text, mbox, link).
Subject: Re: Bug#958218: update-grub fails to process more than one argument
to initrd
Date: Tue, 2 May 2023 12:26:30 +0100
Hi Krzmbrzl,
On Mon, May 01, 2023 at 02:42:41PM +0200, Krzmbrzl wrote:
>Hi everyone,
>
>In this report, there seems to be an indication that this issue was fixed
>with grub2/2.04-5. At least that's how I interpret the auto-generated line
>"No longer marked as found in versions grub2/2.04-5" that was somehow
>triggered by Colin Watson (though when checking the respective mail, it seems
>like that is only about reassignment and not about a resolution).
>
>In either case, I want to report that this issue still exists when having
>grub2-common/2.06 installed (on Ubuntu 22.04), which comes with
>os-prober/1.79.
If you're seeing this in Ubuntu, then you'll need to talk to the
Ubuntu developers. There are two parts to this, one in os-prober
itself and one in GRUB. The os-prober part was fixed in version 1.80
in Debian. The fix in GRUB will be included shortly in Debian - see
https://salsa.debian.org/grub-team/grub/-/commit/358e8faa1329e9ed3467f6f9cacaa781fc34a7b8
I'd suggest you ask the Ubuntu folks to pick that up.
>Furthermore, I can explain exactly where this issue occurs. I have written up
>a question on the Unix&Linux StackExchange about this, that contains the
>explanation: https://unix.stackexchange.com/q/744624/203826
ACK.
--
Steve McIntyre, Cambridge, UK. [email protected]
Mature Sporty Personal
More Innovation More Adult
A Man in Dandism
Powered Midship Specialty