Boot, Limine and LUKS on Omarchy 4.x

Omarchy has used Limine as its bootloader since 2.0. On 4.x the chain is: firmware, Limine menu, a unified kernel image (UKI) on the EFI system partition, an initramfs that unlocks LUKS, then Btrfs root. Everything on this page was checked against the v4.0.4 source tree and against issues filed between August and September 2026.

Status on 4.0.4

For a plain single-disk install from the ISO, this subsystem is boring in the good way: you get a themed menu, a snapshot submenu, a Plymouth splash, and a LUKS prompt. The failures cluster in five places, and almost none of them are hardware faults:

• Config layering. /etc/default/limine has the highest priority of every config layer. One stale value there overrides /etc/kernel/cmdline and every drop-in, silently.
• Unusual firmware. Old UEFI implementations, Apple firmware, and American Megatrends boards each break a different assumption.
• Unusual storage. Software RAID, Intel RST, dual NVMe with two ESPs.
• Dual boot. The documented limine-scan step and the second-ESP layout both have open bugs.
• The 4.0.4 kernel switch. Making linux-omarchy the default boot entry moved the UKI filename, which surprised anything pinned to the old path.

The component tracker counts 692 issues matching boot, Limine, LUKS, UEFI and kernel-panic terms, 387 of them still open, so treat the table below as the sharp edges rather than an exhaustive list.

What Omarchy does automatically

Omarchy installs limine-mkinitcpio-hook and drives everything through limine-entry-tool drop-ins rather than hand-editing boot entries.

The packaged drop-in /etc/limine-entry-tool.d/omarchy-defaults.conf sets the UKI name to omarchy, turns on ENABLE_LIMINE_FALLBACK and FIND_BOOTLOADERS, keeps at most six snapshot entries, and appends a quiet kernel command line: quiet splash loglevel=0 systemd.show_status=false rd.udev.log_level=0 vt.global_cursor_default=0. It also appends initramfs_async=0, with a comment explaining why: kernel 7.1 unpacks the initramfs asynchronously, plymouthd then cannot read /proc/cmdline, and an encrypted machine gets a bare text LUKS prompt instead of the themed one.

BOOT_ORDER in that file reads linux-t2, linux-omarchy, linux-omarchy-*, *, *fallback, Snapshots. T2 Macs stay on their own kernel on purpose; everything else prefers the Omarchy kernel.

Hardware quirks write their own drop-ins during install. install/hardware/apple/fix-t2.sh adds intel_iommu=on iommu=pt pm_async=off mem_sleep_default=deep for T2 Macs along with the linux-t2 kernel; install/hardware/intel/fred.sh adds fred=on on Panther Lake; install/hardware/asus/fix-asus-ptl-b9406-display.sh adds xe.enable_panel_replay=0 on the ExpertBook B9406, and fix-asus-ptl-display-backlight.sh adds xe.enable_dpcd_backlight=1 on that model and the Zenbook UX5406AA. After install, omarchy-hibernation-setup writes resume.conf (and rtc-alarm.conf on s2idle machines) into the same directory, and omarchy-upgrade-to-quattro writes root= into /etc/default/limine when it finds the effective command line has none.

Updates carry migrations that rebuild the boot image when they must. Migration 1786482992 compares the running /proc/cmdline against the defaults drop-in and runs limine-mkinitcpio if parameters are missing. Migration 1789325478, the one that landed in 4.0.4, installs linux-omarchy, rewrites BOOT_ORDER, rebuilds that kernel’s entry, then verifies with limine-entry-tool --tree and refuses to mark itself complete if the entry is absent. It leaves the old kernel installed on purpose so you have something to fall back to.

Snapshots come from install/config/snapper.sh, which writes a root config limited to five snapshots with the timeline disabled, then enables snapper-cleanup.timer and limine-snapper-sync.service. omarchy-snapshot create runs before updates and labels each snapshot with the Omarchy version.

Two commands are worth knowing. omarchy-refresh-limine moves /boot/limine.conf to .bak, copies the packaged default over it, then runs limine-update and limine-snapper-sync. omarchy-setup-direct-boot adds or removes an efibootmgr entry pointing straight at the UKI, and refuses to run on American Megatrends and Apple firmware.

Known problems

Issue	Models affected	Status	Fixed in
#11878 stale PARTUUID in /etc/default/limine kills the LUKS prompt	LUKS plus Btrfs installs moved to a new drive	open	not yet
#9826 root= missing from the UKI command line	bare 3.x installs over a hand-made Limine config	open	not yet
#8047 btrfs-overlayfs blocks limine-snapper-restore	any Btrfs install	open	not yet
#6629 limine-snapper-sync inactive, no snapshot entries	fresh installs, reported from 3.8.4	open	not yet
#12145 direct boot keeps booting the old kernel after 4.0.4	any with direct boot enabled	open	not yet
#12143 UKI chainload panic, EFI_INVALID_PARAMETER	ThinkPad Yoga 11e, older UEFI	open	not yet
#10945 Secure Boot Violation after an update that touches Limine	sbctl custom-key setups	open	not yet
#11526 non-AVX2 CPU left unbootable mid-update	Celeron N4000 and similar, reported from 3.8.5	open	not yet
#7989 leftover archinstall entry panics as the default	archinstall-based installs upgraded to Quattro	open	not yet
#7867 second ESP created, Windows entry dropped on refresh	dual-boot with Windows	open	not yet
#7906 limine-scan writes an uppercase Windows path that panics	dual-boot with Windows	open	not yet
#11997 no BOOTX64.EFI, invisible in the Mac boot picker	Intel Macs	open	not yet
#11101 limine-install fails on RAID or Intel RST ESP	software RAID, Intel RST	open	not yet
#8814 first boot halts in Limine, no direct UKI entry	dual-NVMe, dual-ESP	open	not yet
#11442 mouse movement cancels the autoboot countdown	desktops with a USB mouse	open	not yet
#12096 removing hibernation leaves resume= in the UKI	hibernation users	open	not yet

A few of these are worth a sentence more. In #11878 the reporter’s boot dropped to an emergency shell with no passphrase prompt at all, because a PARTUUID in /etc/default/limine outranked the correct one in /etc/kernel/cmdline. The reporter first read it as an orphaned value, then corrected himself: it was the LUKS partition of an external drive the install had been migrated from, so a check against blkid alone would have passed it. In #9826 the hook’s argument count turned out to be fine; the machine was a bare 3.x install over a hand-made Limine config, which left an empty KERNEL_CMDLINE[default]+="" line that switches off the /etc/kernel/cmdline fallback, and the reporter confirmed that appending root= to /etc/default/limine fixed it. In #10945 a second reporter hit the lockout on a combined 4.0.2 to 4.0.3 update, not only a Limine-only one, and traced it to the installer’s 99-omarchy-limine.hook copying the unsigned binary over the signed one. In #11526 the reporter, on 3.8.5, states that limine-mkinitcpio-hook 1.38.0-1.1 runs on the CPU; the gap is that the update delivering it needs the older binary to rebuild the UKI first. In #11442 the reporter points out that omarchy debug is useless for this class of bug, since the hang happens in Limine before the kernel starts. In #12143 the reporter got a native-Linux entry generated with ENABLE_UKI=no, but says openly that a full boot was not confirmed, so treat that one as a lead rather than a fix.

The 4.0.4 kernel switch has its own tail. #12187 reports a hybrid NVIDIA laptop freezing seconds after login on linux-omarchy while the stock kernel is fine, on a machine carrying prebuilt nvidia-open rather than the DKMS package Omarchy installs.

Where 3.x differed: the boot failures were mostly install-time. In #4152, January 2026, NVIDIA machines finished a 3.3.x install with only an EFI fallback entry in Limine, and @ryanrhughes closed it against the v3.3.2 release. The 4.x failures are config-layer and upgrade-path failures instead.

Fixes that work

Work in this order.

1. Do not reinstall. Limine keeps older entries. Arrow to the previous kernel or a snapshot and boot that first, then fix from a working desktop. With direct boot enabled you must pick Limine from the firmware boot menu to see the snapshot submenu at all.
2. If you land in an emergency shell or get no LUKS prompt, check the effective command line with sudo limine-entry-tool --get-cmdline default and look for root=. Compare any UUID or PARTUUID in /etc/default/limine against /etc/kernel/cmdline and against the device findmnt -no SOURCE / reports, not only against blkid, since a stale value can still name a real partition on another drive. Fix the file, then sudo limine-mkinitcpio.
3. If the menu itself is wrong, run omarchy-refresh-limine. It backs up your current /boot/limine.conf to .bak first. Be aware it replaces the file wholesale, so a hand-added Windows entry does not survive.
4. If snapshots are missing from the menu, check systemctl status limine-snapper-sync.service and confirm a root config exists with sudo snapper --csvout list-configs. If there is none, rerun the packaged install/config/snapper.sh.
5. If the countdown never expires, add mouse: no to the global section of /boot/limine.conf, above the first entry.
6. If Secure Boot refuses Limine after an update, run sudo sbctl verify and, if limine_x64.efi shows as unsigned, sudo sbctl sign -s /boot/EFI/limine/limine_x64.efi before rebooting again.
7. If nothing boots, use the ISO as a rescue system, arch-chroot into the install, repair the config, and run limine-mkinitcpio there.

Report it

Run omarchy debug. It writes /tmp/omarchy-debug.log with inxi -Farz, dmesg, the current boot’s warnings and errors from journalctl, and the full package list, then offers to upload it with a 24 hour expiry. Use omarchy debug --print --no-sudo if you would rather read it first.

For a boot bug that log is often empty of the useful part, because the failure happened before the kernel. Add, by hand: a phone photo of the panic text, efibootmgr -v, sudo limine-entry-tool --tree, ls /boot/EFI/Linux/, and the contents of /etc/default/limine and /etc/limine-entry-tool.d/. Say which kernel entry you picked and whether the other one boots. That last detail is what made #12187 and #12145 readable.

Related

• Kernel panic after update
• You are in emergency mode after update
• Secure Boot violation or will not boot UEFI
• LUKS passphrase not accepted at boot
• Creating a snapshot failed
• Rollback with Snapper and Limine
• Upgrading 3.x to 4 Quattro
• Should you dual boot
• T2 Macs
• Storage and NVMe