OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
omarchy-core high common DISPUTE

Enter the LUKS passphrase when only a Bluetooth keyboard is available

omarchyarchlaptopdesktop
CORRECTED An auditor found the fix wrong or incomplete and rewrote it.
AUDIT NOTE

Checked on this Omarchy 4.0.2-1 workstation (mkinitcpio 41.1-1, cryptsetup 2.8.7-1, kernel 7.1.9) and against upstream. The core claim holds and is stated by upstream itself: the getting-started manual page says "The full-disk encryption won't allow you to enter the password from a Bluetooth keyboard at startup" and tells you to use a 2.4 GHz dongle or a cable, and the manual-installation page repeats it. Both cited learn.omacom.io URLs return HTTP 200 and both contain that text, so they stand. Three things in the record are wrong for Omarchy 4. First, the FIDO2 advice cannot work: systemd-cryptenroll --fido2-device=auto writes a systemd-fido2 LUKS2 token that only systemd-cryptsetup can consume, and Omarchy 4 assigns HOOKS wholesale in /etc/mkinitcpio.conf.d/omarchy_hooks.conf with the busybox encrypt hook. I diffed the two build hooks on this machine: /usr/lib/initcpio/install/sd-encrypt copies libcryptsetup-token-systemd-fido2.so, the tpm2 plugin, libfido2 and the ask-password units, and /usr/lib/initcpio/install/encrypt copies none of them. The Arch wiki Systemd-cryptenroll page says mkinitcpio users must enable the systemd and sd-encrypt hooks for this, and upstream's own hardware-authentication manual page says its Fido2 setup covers sudo "not unlocking your computer". So the record sent a reader to burn a LUKS keyslot for no effect. Second, the AutoEnable=true edit contradicts Omarchy's deliberate design: /usr/share/omarchy/install/hardware/bluetooth.sh says AutoEnable stays at its stock default on purpose and /usr/share/omarchy/migrations/1786380259.sh reverts an AutoEnable=false line, because the power state lives in the rfkill soft block behind omarchy-bluetooth-power. bluetooth.service is already enabled by the installer (confirmed systemctl is-enabled bluetooth.service returns enabled here). Third, /dev/nvme0n1p2 is a guess. The real device is in /proc/cmdline, which reads cryptdevice=PARTUUID=...:root on this box. The SDDM half turned out to be right in a way the record did not explain: Omarchy 4 does ship and enable SDDM (sddm 0.21.0-7 active, /usr/share/omarchy/install/login/sddm.sh), but on an encrypted install the ISO writes /etc/sddm.conf.d/autologin.conf, so there is no greeter prompt and the login-screen symptom only applies to unencrypted installs. I rewrote cause, symptom, fix, danger and verify to say plainly that a Bluetooth keyboard cannot work here at all, and to keep the one no-typing route the shipped hook does support, the cryptkey= keyfile, flagged as a security trade. Dropped the GitHub issue search URL, which is a live query rather than a page and asserts nothing checkable. Not exercised: I did not rebuild an initramfs, did not run systemd-cryptenroll, did not boot with a Bluetooth keyboard, and could not list /boot because it is a vfat ESP mounted dmask=0077. Severity high and frequency common both still fit: the machine will not boot until a wired keyboard is found, but no data is at risk.

Applied by hand after the merge on 2026-09-13: the verdict's fix wrote sudo omarchy-bluetooth-power on, and sudo strips OMARCHY_PATH, so every omarchy subcommand then fails with find: '/themes/': No such file or directory. The lint caught it. /usr/share/omarchy/bin/omarchy-bluetooth-power calls rfkill unblock bluetooth itself and takes no sudo, and the sibling record bluetoothctl-no-default-controller, corrected the same day, already called it without.

The cause above was rewritten on 2026-09-13 to match this note; the fix was corrected by the audit itself.
RISK

systemd-cryptenroll writes to the LUKS header. On a stock Omarchy 4 it buys you nothing at boot, because the encrypt hook cannot use the token it creates, so the only thing it reliably does is consume a keyslot. If you run it anyway, back the header up first and confirm a passphrase keyslot survives:


sudo cryptsetup luksHeaderBackup /dev/disk/by-partuuid/<uuid> \
  --header-backup-file ~/luks-header.img
sudo cryptsetup luksDump /dev/disk/by-partuuid/<uuid>

Never remove the passphrase keyslot. It is the only one the shipped initramfs can use, and removing it locks you out of the disk permanently. If an initramfs or cmdline change does leave the machine unbootable, recover by picking an older entry from Limine's Snapshots submenu at boot, which /etc/limine-entry-tool.d/omarchy-defaults.conf enables. There is no fallback UKI to fall back to: MKINITCPIO_FALLBACK is unset and limine-mkinitcpio-install deletes a fallback UKI when it is.

Symptom

A fresh install boots to the full-disk-encryption passphrase prompt and the Bluetooth keyboard is completely unresponsive, with no characters appearing. The same keyboard works fine once the desktop is up. On an install made without encryption, the same dead-keyboard window is also reported at the SDDM greeter.

Cause

The passphrase prompt runs from the initramfs, long before userspace exists. Omarchy 4 builds that initramfs with the busybox encrypt hook: HOOKS are assigned wholesale in /etc/mkinitcpio.conf.d/omarchy_hooks.conf as base udev plymouth keyboard autodetect microcode modconf kms keymap consolefont block encrypt filesystems fsck btrfs-overlayfs. The keyboard hook it pairs with adds only USB host controllers, usbhid, and the HID and serio input modules (/usr/lib/initcpio/install/keyboard). Nothing in that image runs bluetoothd, and a paired Bluetooth keyboard needs its link keys from /var/lib/bluetooth, which sits on the encrypted root that has not been unlocked yet. The dependency is circular, so adding modules or firmware to the initramfs cannot fix it. Omarchy's own manual states the limit as a prerequisite and tells you to use a wired or 2.4 GHz keyboard.

A separate problem can appear at the SDDM greeter on an install made **without** encryption, where the adapter has not powered up and reconnected the keyboard before the greeter draws. On an encrypted install the ISO writes /etc/sddm.conf.d/autologin.conf, so there is no greeter prompt at all and that second problem does not arise.

Fix

## Use a wired or 2.4 GHz keyboard for the passphrase prompt

This is Omarchy's own answer, and on a stock install it is the only one that works. A 2.4 GHz dongle keyboard enumerates as a plain USB HID device, which the keyboard hook already covers. A Bluetooth keyboard does not, and cannot be made to.

Confirm which device the prompt is unlocking before you change anything. Do not copy a device path out of a guide, read it off this machine:


cat /proc/cmdline | tr ' ' '\n' | grep cryptdevice
lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS

On an Omarchy 4 install that prints something of the form cryptdevice=PARTUUID=<uuid>:root. Use /dev/disk/by-partuuid/<uuid> in any command below.

## FIDO2 and TPM2 unlock do not work on a stock Omarchy 4

Do not reach for systemd-cryptenroll here. It writes a systemd-fido2 or systemd-tpm2 LUKS2 token, and only systemd-cryptsetup, which the sd-encrypt hook installs, can use one. Omarchy 4 ships encrypt, not sd-encrypt. Compare the two build hooks and the difference is explicit: /usr/lib/initcpio/install/sd-encrypt copies /usr/lib/cryptsetup/libcryptsetup-token-systemd-fido2.so, the TPM2 token plugin, libfido2, the TPM character-device modules and the systemd ask-password units into the image, while /usr/lib/initcpio/install/encrypt copies none of them. Enrolling a token on a stock Omarchy 4 therefore spends a keyslot and changes nothing at boot. Omarchy's manual says the same thing from the other end: its FIDO2 setup covers sudo and authorization prompts, "not unlocking your computer".

Check which hook you actually have before believing any guide on this:


grep -o 'HOOKS=([^)]*)' /etc/mkinitcpio.conf.d/omarchy_hooks.conf

Switching to sd-encrypt means adding a later-sorting drop-in (mkinitcpio concatenates /etc/mkinitcpio.conf.d/*.conf in version-sorted order, so a file such as zz-user-hooks.conf overrides omarchy_hooks.conf), adding the systemd hook, changing the kernel cmdline in /etc/kernel/cmdline from cryptdevice= to rd.luks.name=, and rebuilding. Get any one of those out of step and the machine stops at a busybox prompt. It is not worth it to avoid plugging in a cable, and it is not what this record recommends.

## The one no-typing route the shipped hook does support

The busybox encrypt hook reads cryptkey=device:fstype:path from the kernel cmdline and will unlock from a keyfile on a USB stick (/usr/lib/initcpio/hooks/encrypt). That works on a stock Omarchy 4, but it moves the secret onto removable media, so anyone who takes the stick and the laptop has both halves. Treat it as a deliberate trade, not a convenience.

## For the SDDM greeter on an unencrypted install

Do not edit /etc/bluetooth/main.conf. Omarchy leaves AutoEnable at its stock default on purpose (/usr/share/omarchy/install/hardware/bluetooth.sh), and migration /usr/share/omarchy/migrations/1786380259.sh actively reverts an AutoEnable=false line back to a comment, because the power state is held in the rfkill soft block instead. bluetooth.service is already enabled by the installer. If the adapter comes up off, turn it on through Omarchy's own control:


omarchy-bluetooth-power is-on
omarchy-bluetooth-power on

Or use _Update > Hardware > Bluetooth_ in the Omarchy menu (Super + Space).

Verify

Typing on the wired or 2.4 GHz keyboard at the passphrase prompt echoes and the disk unlocks. Confirm the shipped setup is the busybox encrypt hook rather than sd-encrypt, which is what rules FIDO2 and TPM2 out:


grep -o 'HOOKS=([^)]*)' /etc/mkinitcpio.conf.d/omarchy_hooks.conf

Expect ... block encrypt filesystems fsck btrfs-overlayfs. After login, the Bluetooth keyboard should be back:


bluetoothctl devices Connected
omarchy-bluetooth-power is-on

Sources

bluetooth-keyboard-cannot-unlock-luks