### System details

MacBook Pro 13" Late 2013 (A1502, MacBookPro11,1), Haswell/Iris 5100, APPLE SSD SM0256F (Samsung S4LN053X01 AHCI), Omarchy 4.0.3 ISO, live kernel linux-t2 7.2.3-arch1-Watanare-T2-2-t2, boot via Option → EFI Boot, no T2 chip.

### What's wrong?

## Intel MacBook Pro 2013 (A1502) + APPLE SSD SM0256F: live ISO cannot see the disk, then pacstrap dies with Input/output error; GRUB timeout 0 blocks libata.force=noncq

### System details

- Machine: MacBook Pro 13" Late 2013, model A1502, MacBookPro11,1
- CPU / GPU: Intel Haswell, Iris 5100
- Storage: APPLE SSD SM0256F, 256 GB (233.8 GiB usable)
- Controller as seen by the live ISO: `04:00.0 SATA controller: Samsung Electronics Co Ltd S4LN053X01 AHCI SSD Controller (Apple slot) (rev 01)`
- Omarchy ISO: 4.0.3 (`omarchy-4.0.3.iso`)
- Live kernel: `linux-t2` (`7.2.3-arch1-Watanare-T2-2-t2`)
- Boot path: Apple Option menu → orange EFI Boot
- This Mac has no T2 chip. Startup Security Utility from the Intel Mac doc does not exist on this hardware.

macOS Disk Utility reported SMART Status: Verified on this SSD before the Linux attempts. The drive is healthy under macOS.

### What happened

Two failures stacked. Together they made a clean install impossible from this ISO.

**1. Installer aborts at disk selection because the internal SSD is not a block device**

First boots showed only the USB installer in `lsblk` (SanDisk Cruzer Blade, ~15 GB). The Omarchy configurator then printed:

```
Let's select where to install Omarchy...
no options provided, see `gum choose --help`
Aborted installation
```

That matches `disk_form()` in the ISO configurator: it lists `/dev/sd*`, `/dev/nvme*`, etc., then excludes the archiso boot disk. With only the USB present, `gum choose` gets an empty list and aborts.

`dmesg` on those boots:

```
ahci 0000:04:00.0: AHCI vers 0001.0300, 32 command slots, 6 Gbps, SATA mode
ahci 0000:04:00.0: 1/1 ports implemented (port mask 0x1)
ata1: SATA link up 6.0 Gbps
ata1.00: qc timeout after 5000 msecs
ata1.00: failed to IDENTIFY (I/O error, err_mask=0x4)
ata1: SATA link up 6.0 Gbps
ata1.00: qc timeout after 10000 msecs
ata1.00: failed to IDENTIFY (I/O error, err_mask=0x4)
ata1: limiting SATA link speed to 3.0 Gbps
ata1: SATA link down
```

PCI sees the Samsung Apple-slot AHCI controller. libata never finishes IDENTIFY, so there is no `/dev/sdX` for the internal drive.

This is the known Apple PCIe AHCI SSD behavior on 2013-2015 MacBook Pros under Linux. The usual workaround is `libata.force=noncq` (sometimes `libata.force=noncq,3.0G`) on the kernel command line, before the AHCI probe.

**2. I could not pass that parameter on this ISO**

`/run/archiso/bootmnt/boot/grub/grub.cfg` is the live boot config. The `linux` line is the T2 kernel with `quiet splash xe.enable_panel_replay=0 initramfs_async=0`.

Problems:

- GRUB timeout is effectively 0. Esc, Shift, and `e` never opened the editor. Boot went straight into the keyboard-layout wizard.
- `mount -o remount,rw /run/archiso/bootmnt` failed: the ISO filesystem is write-protected.
- So there is no in-place way to add `libata.force=noncq` on a stock 4.0.3 stick.

The Intel Mac page documents Broadcom Wi-Fi, SPI keyboard, and NVMe suspend. It does not mention this AHCI IDENTIFY failure or how to edit live kernel args.

**3. After a macOS wipe the disk appeared, then pacstrap failed on writes**

From OS X Recovery Disk Utility I erased the 251 GB APPLE SSD (GUID, Mac OS Extended). A later live boot then listed:

```
/dev/sda (233.8G) - ATA APPLE SSD SM0256F [vfat, hfsplus]
```

The installer accepted that disk. pacstrap then died while writing into the new root:

```
error: could not create directory /mnt/var/lib/pacman/local/expat-2.8.4-1/: Input/output error
==> ERROR: Failed to install packages to new root
```

A later retry got further (libnghttp2, libnghttp3, nettle, leancrypto) and died the same way on `gnutls`:

```
error: could not create directory /mnt/var/lib/pacman/local/gnutls-3.8.13-2/: Input/output error
```

Same class of failure, different package. This is not a missing package on the offline mirror. The target SSD rejects writes under this kernel.

Setting `/sys/block/sda/device/queue_depth` to `1` after the disk was already probed was not enough. `wipefs -a /dev/sda` then failed with:

```
wipefs: /dev/sda: cannot flush modified buffers: Input/output error
```

One retry also stopped earlier with:

```
Luks or LvmOnLuks encryption require partitions to be defined
```

That looked like leftover state from the failed pacstrap, not the root cause.

### What I expected

On a machine the Mac support page claims to handle, the live ISO should either:

1. expose the internal Apple AHCI SSD as a block device without extra kernel args, or
2. document and allow `libata.force=noncq` at boot (non-zero GRUB timeout, or a second menu entry that already has that parameter).

Once the disk is visible, pacstrap should complete without Input/output errors on a drive that macOS considers SMART-verified.

### Steps to reproduce

1. Flash `omarchy-4.0.3.iso` with balenaEtcher.
2. On a 2013 13" MacBook Pro A1502 with the original SM0256F, hold Option and boot the orange EFI Boot entry.
3. Observe that GRUB does not stay long enough to edit the `linux` line.
4. Let the installer run to "Let's select where to install Omarchy".
5. If the SSD was not recently reset under macOS, `lsblk` shows only the USB and the installer aborts with `no options provided`.
6. `dmesg` shows `ata1.00: failed to IDENTIFY (I/O error, err_mask=0x4)` and then `SATA link down`.
7. Wipe the SSD from OS X Recovery, reboot the same ISO. The disk may now appear as `/dev/sda`.
8. Continue a full-disk install. pacstrap fails with Input/output error while creating directories under `/mnt/var/lib/pacman/local/`.

### What I tried

- `e` / Esc / Shift at GRUB: no editor
- remounting the ISO bootfs read-write: write-protected
- OS X Disk Utility erase of the 251 GB Apple SSD
- `echo 1 > /sys/block/sda/device/queue_depth` before retry
- `wipefs -a` on `/dev/sda`, `/dev/sda1`, `/dev/sda2`: I/O error
- `./.automated_script.sh` several times after the first abort
- install with encryption still selected after a partial layout existed

I have not yet been able to boot this ISO with `libata.force=noncq` on the kernel command line. Next attempt is the same 4.0.3 ISO via Ventoy from another machine, so I can pass that parameter before AHCI probes the SM0256F.

### Suggested direction (not a patch)

- Add a short GRUB timeout (or an "Omarchy (Apple AHCI SSD)" entry) that includes `libata.force=noncq` on the live ISO.
- On Mac detection, if `lspci` shows the Samsung `S4LN053X01` Apple-slot AHCI controller and `lsblk` has no non-USB disk, print the IDENTIFY failure and the needed kernel parameter instead of an empty `gum choose`.
- Mention this 2013-2015 Apple AHCI SSD case on the Intel Mac support page. The current auto-fixes (Broadcom, SPI keyboard, NVMe suspend) do not cover it.
- Worth checking whether shipping `linux-t2` on a pre-T2 2013 Mac changes AHCI timing. I cannot isolate that without a non-T2 live kernel.

I can rerun with `libata.force=noncq` and report whether IDENTIFY and pacstrap both succeed.