Fix 'Call to Hibernate failed: No such file or directory' with a swapfile in /home
The ProtectHome=yes diagnosis is a genuine, well-known logind gotcha and the Btrfs path is correct (btrfs filesystem mkswapfile -s SIZE file is real per btrfs-filesystem.8, which also states "A swapfile must be created in a specific way: NOCOW and preallocated"). The ext4/XFS path is the weak spot: fallocate creates unwritten/preallocated extents, and on XFS swapon rejects the resulting file ("swapfile has holes"). The modern, filesystem-agnostic way is mkswap --size --file, which allocates correctly. The record also never adds the fstab entry for the ext4/XFS case, so the swapfile disappears on reboot.
Deleting the old swapfile while it is still referenced in /etc/fstab leaves a boot that waits on a missing swap unit. Remove or update the fstab line in the same change.
Symptom
systemctl hibernate fails instantly with Call to Hibernate failed: No such file or directory, even though swapon --show clearly lists my swapfile and it is bigger than RAM.
Cause
systemd-logind runs with ProtectHome=yes, so it cannot see files under /home, /root or /run/user. Its pre-hibernation swap-size check therefore fails to stat the swapfile and reports ENOENT.
Fix
sudo swapoff /home/$USER/swapfile
sudo rm /home/$USER/swapfile
Recreate at the top level. Use mkswap --size --file rather than fallocate, because fallocate produces unwritten extents that swapon refuses on XFS:
sudo mkswap -U clear --size 32G --file /swapfile
sudo chmod 0600 /swapfile
sudo swapon /swapfile
(On older util-linux without --file, use dd if=/dev/zero of=/swapfile bs=1M count=32768 status=progress, then chmod 600, mkswap, swapon. Do not use fallocate.)
On Btrfs, use a NOCOW subvolume, as Omarchy does:
sudo btrfs subvolume create /swap
sudo chattr +C /swap
sudo btrfs filesystem mkswapfile -s 32g /swap/swapfile
sudo swapon -p 0 /swap/swapfile
Add to /etc/fstab in BOTH cases, or it is gone after reboot:
/swapfile none swap defaults 0 0
# or, Btrfs:
/swap/swapfile none swap defaults,pri=0 0 0
Then redo resume= / resume_offset= for the new file (see the resume-offset record) and rebuild the initramfs. Sanity-check with swapon --show and systemctl hibernate.
Verify
systemctl hibernate no longer errors out immediately. swapon --show lists the new path and findmnt -no UUID -T /swap/swapfile resolves the backing device.