I haven't rebooted my computer since August 2017, and did some selective apt update/install in the meantime.
Today I restarted, and I got stuck into the initramfs :
Code: Select all
Decompressing Linux... Parsing ELF... No relocation needed... done.
Booting the kernel. Loading, please wait...
Please unlock disk/dev/disk/by-uuid/ff29...5d59 (sdb1): <password>
cryptsetup (sdb1): set up successfully
fsck from util-linux 2.27.1
Usage: mount [-r] [-w] [-o options] [-t type] [-f] [-i] [-n] device directory
Target filesystem doesn't have requested /sbin/init.
mount: No such file or directory
Could not copy file: No such file or directory
No init found. Try passing init= bootarg.
Busybox v1.27.2 (Debian 1:1.27.2-2) built-in shell (ash)
(initramfs) _
In the initramfs, I have been able to manually mount the partition on /root or /mnt.
But when I try to type "exit" or "exec switch_root /mnt /sbin/init", I always get a kernel panic.
For information, I have other partitions (encrypted and non-encrypted) on the same disks, which I am able to manually open and mount correctly also. I don't mention them here because it seems not related to my issue. I saw that /dev/mapper/sdb1 is also created as /dev/dm-0, I don't know why but it seems not related with the issue. I don't have any swap partition and I try to avoid using LVM.
After searching on the Internet, I tried a few things:
- run manually fsck on all the devices with success.
- mount the partitions without the option "-t ext3". I got an error similar to boot sequence: Usage: mount [-r] [-w] [-o options] [-t type] [-f] [-i] [-n] device directory. So I tried to find the script which does a "mount" command in /etc/initramfs-tools/, but I couldn't find it.
- rename the encrypted partitions differently than their device name in /etc/crypttab and /etc/fstab ("crypto-root" instead of "sdb1"). I managed to have the boot sequence creating the mapped device with this name, but it didn't help.
- use "UUID=xxx" instead of device names in crypttab, in fstab and in grub (root=/dev/mapper/crypto-root, root=/dev/dm-0, or root=UUID=xxx). It didn't help either.
- copy scripts/local-top/ and hooks/cryptroot, from /usr/share/initramfs-tools to /etc/initramfs-tools/, since those scripts dated from 2010 whereas the /usr/share scripts dated from Sept 14th, 2017. However, I haven't copied the whole content from /usr/share/initramfs-tools, since only the cryptroot script was initially set in /etc when it used to work properly.
- mount /sys, /proc, /dev and the deciphered root partition, chroot it and run "update-initramfs -u" and "update-grub". It generates.
- run the recovery-mode sequence from grub (option "single"), it gives the following detailed output :
Code: Select all
Begin: Running /scripts/init-premount... done.
Begin: Mounting root file system...
Begin: Running /scripts/local-top...
Please unlock disk crypto-root: <password>
[ ] bio: create slab <bio-1> at 1
cryptsetup (crypto-root): set up successfully
Begin: Assembling all MD arrays...
Failure: failed to assemble all arrays.
done.
done.
Begin: Running /scripts/local-premount... done.
Begin: Checking file system...
fsck from util-linux 2.27.1
done.
Usage: mount [-r] [-w] [-o options] [-t type] [-f] [-i] [-n] device directory
done.
Target filesystem doesn't have requested /sbin/init.
Begin: Running /scripts/local-bottom... done.
Begin: Running /scripts/init-bottom...
mount: No such file or directory
done.
Could not copy file: No such file or directory
No init found. Try passing init= bootarg.
Busybox v1.27.2 (Debian 1:1.27.2-2) built-in shell (ash)
(initramfs) _
So I am able to mount and chroot all the devices manually, but I am not able to continue the boot sequence manually afterwhile (kernel panic), and I am not able to make initramfs automatically goes further the mount error.
Do you see why the line "Usage: mount" appears at this level of the boot sequence?
How to hack initramfs in order to have it pass the mount command?
Which additional steps may I try?
Thanks for your help.
-Sylvain
