Repository navigation
hibernation remove leaves resume kernel parameters in the UKI (Limine drop-in not removed) #12096
Description
Summary
omarchy hibernation setup writes the resume kernel parameters to /etc/limine-entry-tool.d/resume.conf (and, on s2idle systems, /etc/limine-entry-tool.d/rtc-alarm.conf), but omarchy hibernation remove never deletes those drop-ins. It only removes the mkinitcpio hook (/etc/mkinitcpio.conf.d/omarchy_resume.conf) and then runs limine-mkinitcpio.
As a result, after removal the rebuilt UKI still embeds resume=/dev/mapper/root resume_offset=…, pointing at a swapfile that no longer exists. The kernel can't find a resume image and falls through to a normal boot, but the boot configuration is left stale — remove is not symmetric with setup.
Steps to reproduce
-
omarchy hibernation setup -
Confirm both drop-ins exist:
/etc/mkinitcpio.conf.d/omarchy_resume.conf/etc/limine-entry-tool.d/resume.conf(containsresume=/resume_offset=)
-
omarchy hibernation remove -
/etc/limine-entry-tool.d/resume.confis still present, and the rebuilt UKI still contains the resume parameters:objcopy -O binary --only-section=.cmdline /boot/EFI/Linux/omarchy_linux.efi /dev/stdout | tr -d '\0' # ... still shows resume=/dev/mapper/root resume_offset=...
Expected
omarchy hibernation remove should remove the Limine drop-ins that setup created, before rebuilding the boot image — mirroring what setup does. Something like:
sudo rm -f /etc/limine-entry-tool.d/resume.conf
sudo rm -f /etc/limine-entry-tool.d/rtc-alarm.conf
next to the existing removal of /etc/mkinitcpio.conf.d/omarchy_resume.conf, followed by the usual limine-mkinitcpio.
Relevant code
omarchy-hibernation-setupdefinesRESUME_DROP_IN="/etc/limine-entry-tool.d/resume.conf"and writes it; it also writes/etc/limine-entry-tool.d/rtc-alarm.confwhen/sys/power/mem_sleepcontains[s2idle].omarchy-hibernation-removeonly referencesMKINITCPIO_CONF(/etc/mkinitcpio.conf.d/omarchy_resume.conf) and never touches/etc/limine-entry-tool.d/.
Environment
- Omarchy 4.0.3-1
limine12.8.0-1,limine-mkinitcpio-hook1.38.0-1.1- Kernel 7.2.3-arch1-3
- Root on btrfs over LUKS, UKI built via
limine-entry-tool
Activity
omarchybot commentedon Oct 12, 2026
This is the same bug as #10037, which was opened two weeks earlier and is still open, so I'm closing this one as a duplicate.
I checked it against
quattro(9480d1b). The mechanism is as you describe.bin/omarchy-hibernation-setupwrites/etc/limine-entry-tool.d/resume.conf, and on s2idle machines/etc/limine-entry-tool.d/rtc-alarm.confas well.bin/omarchy-hibernation-removedeletes the swapfile, the/swapsubvolume, the fstab entry and/etc/mkinitcpio.conf.d/omarchy_resume.conf, but it never touches/etc/limine-entry-tool.d/. So thelimine-mkinitcpioit runs at the end bakesresume=/resume_offset=back into the UKI. #10037 reports the same missing removal with the same reproduction and the same proposed fix, and it also coversrtc-alarm.conf.Two open pull requests already delete
resume.confinomarchy-hibernation-removebefore the rebuild: #13584 and #13768. Choosing between them is up to the maintainer. Neither removesrtc-alarm.conf. That file only holds the fixedrtc_cmos.use_acpi_alarm=1, with no offset that could go stale, so leaving it behind is untidy but not harmful.I checked this by reading the source; nothing needed running. Codex Medium gave a second opinion on whether this is a duplicate. If they turn out to be different bugs, this can be reopened.