ahci: scan all ports; fat: mount first valid FAT32 disk

The kernel previously started only AHCI port 0, so any launch where the
firmware/QEMU put something else on port 0 (e.g. the ESP from an implicit
-drive file=fat:rw:esp) left the real disk unnoticed: 'unsupported sector
size 0' and no mount.

- ahci.rs: enumerate CAP.NP/PI ports, start every port that links up
  (empty ports settle at DET=0 and are skipped quickly), keep up to 8
  disks, and expose disk_count()/select() for the filesystem
- fat.rs: try each disk in port order and mount the first valid FAT32
  volume; non-FAT disks are logged and skipped
- scripts/boot.sh: attach the data disk on ide.0 and the ESP on ide.1
  (DISK_IMG env var, default /tmp/disk.img), matching the Makefile
- Verified with the disk on port 0 (Makefile style), port 5 (old-style
  command line), and with the ESP on port 0; fsck.fat stays clean

1 files changed, 5 insertions, 1 deletions

diff --git a/CONTEXT.md b/CONTEXT.md
index 24a4b1e..aa9385b 100644
--- a/CONTEXT.md
+++ b/CONTEXT.md
@@ -170,7 +170,11 @@ Single Rust crate (`x86_64-unknown-uefi` target).  Module breakdown:
 ✅ **AHCI + FAT32 (persistent disk)** — a SATA AHCI driver (polling DMA,
   no interrupts) plus a FAT32 filesystem, mounted as `/disk`; every shell
   command works on disk paths: `ls`/`cat`/`hexdump`/`echo >`/`>>`/
-  `mkdir`/`touch`/`rm`/`edit`.  Whole-FAT cache with write-back to both
+  `mkdir`/`touch`/`rm`/`edit`.  The driver scans **all** implemented
+  ports (CAP.NP/PI) and starts every port with a device; `fat::init`
+  tries each disk in port order and mounts the first valid FAT32
+  volume — so the kernel finds the disk no matter which AHCI port the
+  firmware/QEMU put it on (real hardware included).  Whole-FAT cache with write-back to both
   FAT copies, checksum-validated LFN read/write, multi-cluster files,
   space-padded `.`/`..` dot entries (`..` → cluster 0 under the root),
   FSInfo free-count maintenance (fsck-clean after a session, verified