## Summary

After updating to Omarchy 4.0.4, which installs `linux-omarchy` and makes it the default Limine entry, my machine hard-resets about 2 seconds into every boot. The CPU logs a fatal machine-check error each time. The stock Arch kernel (`linux` 7.2.3-arch1-3) boots fine on the same install, so the problem appears to be specific to `linux-omarchy` 7.2.5-3.

## System

- **Omarchy:** 4.0.4-1 (updated from 4.0.2-1 on 2026-10-10)
- **CPU:** AMD Ryzen 7 5800X3D (family 19h model 21h, Zen 3), microcode 0xa201213, amd-ucode 20260810-2
- **GPU:** AMD Radeon RX 9070 / 9070 XT (Navi 48)
- **Motherboard:** Gigabyte X570 AORUS ELITE, BIOS F40 (2025-10-28)
- **RAM:** 16 GB
- **Disk:** NVMe, LUKS + btrfs, Limine + UKI

## Steps to reproduce

1. Run `omarchy update` from 4.0.2. Migration `1789325478` installs `linux-omarchy` 7.2.5-3 and sets `BOOT_ORDER="linux-omarchy, ..."`.
2. Reboot into the default `linux-omarchy` entry.
3. Early in userspace (around 18–19 s monotonic, right after modules like `snd_hda`, `amd_atl` and `intel_rapl_common` load and `/boot` and `/home` mount), the machine resets with no panic or oops in the journal.

This reproduced on 2 out of 2 boots with `linux-omarchy`.

## Expected

The system boots normally, as it does with `linux` 7.1.9-arch1-2 and 7.2.3-arch1-3.

## Evidence

The machine-check banks persist across the warm reset, so the following boot reports the error that killed the previous one:

```
# reported at start of boot after the 1st crash (linux-omarchy 7.2.5-3)
[Hardware Error]: System Fatal error.
[Hardware Error]: CPU:6 (19:21:2) MC5_STATUS[-|UE|MiscV|AddrV|PCC|TCC|SyndV|-|-|-]: 0xbea0000000000108
[Hardware Error]: Error Addr: 0x01ffffffc08073c4
[Hardware Error]: IPID: 0x000500b000000000, Syndrome: 0x000000004d000000
[Hardware Error]: Execution Unit Ext. Error Code: 0
[Hardware Error]: cache level: RESV, tx: GEN, mem-tx: GEN

# reported at start of boot after the 2nd crash
[Hardware Error]: System Fatal error.
[Hardware Error]: CPU:12 (19:21:2) MC5_STATUS[-|UE|MiscV|AddrV|PCC|TCC|SyndV|-|-|-]: 0xbea0000000000108
```

- MC5 is the Execution Unit bank, and extended error code 0 is a core watchdog timeout: a core stopped retiring instructions.
- Each crash happened on a different core (6 and 12) but at the same point in boot. That points to a kernel or code-path trigger rather than one bad core.
- The error address `0x...ffffffc08073c4` falls in the kernel module mapping range.
- In 13 earlier boots on `linux` 7.1.9 and the clean boot on `linux` 7.2.3, no Hardware Error was ever logged. The only other boot that reported one was the 7.1.9 recovery boot, and it was reporting the previous crash.
- `linux-omarchy` is built with GCC 16.2.1, generic x86-64 (`CONFIG_X86_NATIVE_CPU` not set), `PREEMPT_DYNAMIC`, `HZ=1000`, `AMD_PSTATE_DEFAULT_MODE=3`.

## Workaround

Pick the `linux` entry in Limine. To make it permanent, set `BOOT_ORDER="linux, linux-omarchy, linux-omarchy-*, *, *fallback, Snapshots"` in `/etc/default/limine` and run `sudo limine-update`.

Since the migration keeps the old kernel installed, users can recover this way. But a machine that resets this early gives no hint as to why, so you may want to consider a boot-success check or fallback before making `linux-omarchy` the default.