### System details

Framework Laptop 13 (AMD Ryzen AI 300), Omarchy 3.5.0

### What's wrong?

Below is my debugging with Claude. After applying the fix, it works again.

----------------------------------------------------------------------------------------------------------------------
# Generated with Claude

## Hibernate resume broken after kernel update: `resume_offset=` is empty in boot config

### Description

Hibernate resume has been failing consistently since the kernel was updated on Apr 3 (6.19.10-arch1-1). The system hibernates successfully but does a fresh boot instead of resuming — the hibernate image is discarded every time.

The root cause is that `/etc/limine-entry-tool.d/resume.conf` contains an empty `resume_offset=`:

```
KERNEL_CMDLINE[default]+="resume=/dev/mapper/root resume_offset="
```

This propagates to `/boot/limine.conf`:

```
cmdline: ... resume=/dev/mapper/root resume_offset=
```

Without the offset, the initramfs `resume` hook cannot locate the hibernate image on the btrfs swapfile. The system boots normally, and `swapon` overwrites the hibernate image:

```
swapon: /swap/swapfile: software suspend data detected. Rewriting the swap signature.
```

### Evidence

Hibernate worked reliably from early March through Apr 3 (28 successful hibernate/resume cycles in journalctl). Since the kernel update on Apr 3, every hibernate resume has failed (4 consecutive failures: Apr 5, 7, 8, 9):

```
# Every boot since Apr 5 shows this pattern:
systemd-hibernate-resume[704]: Reported hibernation image: ID=arch kernel=6.19.10-arch1-1 UUID=068db438-b09d-4266-8fe1-edd9cf5cc6a9 offset=45139143
systemd-hibernate-resume[704]: Successfully cleared HibernateLocation EFI variable.
swapon[725]: swapon: /swap/swapfile: software suspend data detected. Rewriting the swap signature.
```

`systemd-hibernate-resume` knows the correct offset (45139143) from the EFI variable, but by the time it runs, the filesystem is already mounted — it's too late. The resume must happen in the initramfs.

### Setup

- Framework Laptop 13 (AMD Ryzen AI 300)
- LUKS-encrypted btrfs root (`/dev/mapper/root`)
- Swap: `/swap/swapfile` (30.6G btrfs swapfile)
- Kernel: 6.19.10-arch1-1 (installed Apr 3)
- mkinitcpio 40-4
- Resume hook is present via `/etc/mkinitcpio.conf.d/omarchy_resume.conf`

### Likely cause

It looks like `omarchy-hibernation-setup` (or `limine-entry-tool` during kernel update) regenerated the boot config but failed to compute/preserve the `resume_offset` value from `btrfs inspect-internal map-swapfile`. The correct offset is **45139143** (as seen in the EFI HibernateLocation variable that systemd writes during hibernate).

### Fix

Manually setting the offset and regenerating the boot config fixed the issue:

```bash
OFFSET=$(sudo btrfs inspect-internal map-swapfile -r /swap/swapfile)
sudo sed -i "s/resume_offset=.*\"/resume_offset=${OFFSET}\"/" /etc/limine-entry-tool.d/resume.conf
sudo limine-mkinitcpio && sudo limine-update
```

After this, hibernate and resume works correctly again. Confirmed via journalctl — same boot ID preserved across hibernate/resume:

```
# Hibernate entry (08:29)
Apr 09 08:29:13 systemd-logind[1094]: The system will hibernate now!
Apr 09 08:29:13 kernel: PM: hibernation: hibernation entry

# Successful resume (08:32) — same boot, no "Rewriting the swap signature"
Apr 09 08:32:08 kernel: PM: hibernation: Allocated 12580444 kbytes in 2.10 seconds (5990.68 MB/s)
Apr 09 08:32:08 kernel: PM: hibernation: hibernation exit
Apr 09 08:32:08 systemd-sleep[270380]: System returned from sleep operation 'hibernate'.
Apr 09 08:32:08 systemd-logind[1094]: Operation 'hibernate' finished.
```

### Underlying bug

The `resume_offset` value is lost when the boot config is regenerated during a kernel update. `omarchy-hibernation-setup` or `limine-entry-tool` should preserve or recompute the offset (from `btrfs inspect-internal map-swapfile -r /swap/swapfile`) when rebuilding boot entries.

### Additional note

There is also a stale `/swapfile` entry in `/etc/fstab` that fails every boot with `BTRFS warning: swapfile must not be copy-on-write`. This is separate from the working `/swap/swapfile` but causes noisy errors on every boot.