No LUKS prompt and an emergency shell after moving Omarchy to a new drive (stale PARTUUID in /etc/default/limine)
#11878 and its correction comment support the symptom, the migration story, the snapshot entries that still boot, and the fix of editing /etc/default/limine and rerunning limine-mkinitcpio. /usr/lib/limine/limine-common-functions load_config() on this machine loads /usr/share/limine-entry-tool.d, /etc/limine-entry-tool.conf, /etc/limine-entry-tool.d and then /etc/default/limine with the quoted '(highest priority)' comment. This machine's /etc/default/limine carries KERNEL_CMDLINE[default]+="cryptdevice=PARTUUID=...:root root=/dev/mapper/root ...", which matches the record. The cause repeats the reporter's framing that the higher-priority file overrode a correct /etc/kernel/cmdline. The packaged README and /etc/limine-entry-tool.conf both say the tool reads /etc/kernel/cmdline or /proc/cmdline only when KERNEL_CMDLINE is unset, and that += ignores both. So /etc/kernel/cmdline is a fallback that Omarchy never reaches, not a lower layer. The cause is rewritten to say that, and to say that the reporter could not find what wrote the stale value. The fix and verify are sound and kept. Not exercised: nothing was rebuilt.
Editing the root or cryptdevice parameters wrongly makes every rebuilt entry unbootable. Keep a known-good snapshot entry, and have the Omarchy ISO ready for the first reboot.
Symptom
After cloning or migrating an Omarchy install to a new drive, or after the next update that regenerates the boot entries, the machine never asks for the disk passphrase and lands in the initramfs emergency shell. /dev/mapper contains only control. Older Btrfs snapshot entries in the Limine menu still boot, which makes it look like a kernel regression. Running limine-mkinitcpio again reproduces the bad entry every time.
Cause
limine-entry-tool builds the command line from four config layers and reads /etc/default/limine last, with the highest priority ("4. Load /etc/default/limine (highest priority)" in /usr/lib/limine/limine-common-functions). On Omarchy its KERNEL_CMDLINE[default]+= line carries cryptdevice=PARTUUID=...:root root=/dev/mapper/root. /etc/kernel/cmdline is only a fallback: the tool reads it, or /proc/cmdline, when KERNEL_CMDLINE is unset (README and /etc/limine-entry-tool.conf), and Omarchy always sets it. A correct value in /etc/kernel/cmdline therefore changes nothing. In the reported case the PARTUUID in /etc/default/limine belonged to the LUKS partition on the old external drive the system had been migrated from. Nothing checks that value against the disk being booted before baking it into the UKI, and the drive was attached, so the PARTUUID was real. Snapshot entries built before the change still carried the right value. What wrote the stale value is not known: the reporter found no Omarchy or Limine script that writes a PARTUUID into that file.
Fix
## Get in once
From the emergency shell, open the real LUKS partition under the mapper name root= expects (Omarchy: root=/dev/mapper/root), then continue booting:
cryptsetup open /dev/nvme0n1p2 root
exit
Or pick an older snapshot entry in Limine.
## Find and correct the stale value
lsblk -o NAME,PARTUUID,FSTYPE,SIZE # the crypto_LUKS partition's PARTUUID
grep -n 'PARTUUID\|root=' /etc/default/limine /etc/limine-entry-tool.d/*.conf /etc/kernel/cmdline 2>/dev/null
Edit /etc/default/limine so cryptdevice=PARTUUID= matches the crypto_LUKS partition on the disk you boot from. Then rebuild:
sudoedit /etc/default/limine
sudo limine-mkinitcpio
Verify
After a reboot the passphrase prompt appears, and grep -o 'cryptdevice=[^ ]*' /proc/cmdline shows the PARTUUID that lsblk -o NAME,PARTUUID,FSTYPE reports for the crypto_LUKS partition.