# USB 32 KiB matrix Round 2 Pi result: USB05_4G and USB01_2G

Status: complete
Result: complete; PASS
Type: evidence
Updated: 2026-08-01
Next Action: none; retain as historical round evidence for the completed matrix
Campaign: 003-complete-bounded-hpt-slow-export-firmware-ticket

Project Map: work/maps/map-smi-firmware-stabilization-and-qualification.md
Workpackage: work/wp/completed/wp-restore-broad-usb-compatibility-baseline.md
Checklist: work/checklists/checklist-stage-1b-four-drive-usb-compatibility-baseline.md
Controller result: work/evidence/usb-32k-round2-controller-usb05-usb01-20260731.md

## Returned identity and filesystem gate

The Raspberry Pi identified both returned drives without ambiguity and kept
them unmounted except for temporary read-only content inspection:

| Canonical label | Pi device | Capacity | Partition identity | FAT UUID |
| --- | --- | ---: | --- | --- |
| `USB05_4G` | `/dev/sda` | 4,026,531,840 bytes | MBR `0xb820a071`; FAT32 LBA `2,048..7,864,319` | `362E-8170` |
| `USB01_2G` | `/dev/sdb` | 2,013,265,920 bytes | MBR `0x43837690`; FAT32 LBA `2,048..3,932,159` | `D45F-70AA` |

Model, revision, VID:PID, capacity, partition table, label, and UUID match the
established specimen identities. `sg_readcap` independently returned final
LBAs `7,864,319` and `3,932,159` with 512-byte logical blocks.

`fsck.fat -n -v` reached the final filesystem sector on both drives, found no
structural corruption, and left each filesystem unchanged. Exit status `1`
is solely the preserved Windows root-directory label versus blank boot-sector
label convention. Both FAT copies are byte-identical:

- `USB05_4G`: each 3,923,456-byte FAT SHA-256
  `DC204E88B1F985E62BBB124DE59341B0D83C9D6489119B27B86117543EDC7476`;
- `USB01_2G`: each 1,961,472-byte FAT SHA-256
  `B6A43EB2B907E1E09788026A4505C71EB1A01D1E5AED2F400829665A6DC0C312`.

The current insertion kernel log contains normal high-speed enumeration,
capacity, partition, and removable-disk attachment events. It contains no
reset, timeout, medium, offline, buffer/block-I/O, sense-key,
descriptor-read, rejecting-I/O, or other error. Both drives are unmounted
after validation.

## Exact output validation

| Output | USB05_4G | USB01_2G |
| --- | --- | --- |
| New HPT CSV | `K2A_USB05.csv`; 442,642 bytes; SHA-256 `CF20ECF08FD1BDBBE3C6A033B25E25BF360BEE4AAA322E835DFC9A68725154A5` | `K2B_USB01.csv`; 443,162 bytes; SHA-256 `9BBEE8EDB69D55B02165F8050744521833639F35D925C6229FFB82DFD37E87D1` |
| Data lines | exact sequence `1..8809`; no missing or duplicate line | exact sequence `1..8817`; no missing or duplicate line |
| Metadata | requested/actual range `301739760..302285220`; 8,809 healthy, 0 degraded, 0 gaps | requested/actual range `301741860..302285280`; 8,736 healthy, 81 degraded, 0 gaps |
| Completion | snapshot `COMPLETE`; `Export completion: COMPLETE; records published: 8809` | snapshot `COMPLETE`; `Export completion: COMPLETE; records published: 8817` |
| New PID | `MainPid 2026-07-31 15-58.txt`; 225 bytes; SHA-256 `6BCB89F2EFACC503B950EF00151A51AF8353C40B53781110C14DDC441737FC3A` | same name, size, and hash |
| Recovery image | 8,388,864 bytes; SHA-256 `F2C11398416DF829078BD960E67D0AE6E54AE1EB8D361A49DB273A6180048E52` | same size and hash |

The 81 degraded Controller B records are the same HPT source-data
classification reported by the controller; they are present in the complete
CSV and are not a USB or filesystem failure.

The replacement image on each drive has title/version
`Version 00-05-00  7520553_SMI_00_05_00.imx`, update version `1`, and SMI
machine mask `1`. Independent read-only validation using the configured
Renesas CRC-32, MSB-first, 32-bit-word input convention passed:

- exact expected size: 256-byte header plus 4,194,304-byte main image plus
  4,194,304-byte audio image;
- header CRC stored/calculated: `0xA5ECD768`;
- main-image CRC stored/calculated: `0xDF597832`;
- audio-image CRC stored/calculated: `0xB2FE8C3B`;
- I/O image length and CRC both zero, as required for SMI.

Both retained WAV files remain exact and valid mono PCM, 44.1 kHz, 16-bit:

- `W00.wav`: 198,566 bytes; declared PCM 198,450 bytes; SHA-256
  `37A9DC665366F06D8081B23ECF88BDB30CF81546C3A10EDA1BE827DE2D006B92`;
- `W01.wav`: 176,516 bytes; declared PCM 176,400 bytes; SHA-256
  `E22FC8FFF2CF68878C56EA87660CA2325E0FBF04CF091AAEB112A7F274B1B2F7`.

The Round 1 CSVs remain readable with their exact prior hashes:
`K1B_USB05.csv` is
`8F4615098398B83EEE4473AFA72682BFFC231B524FB8E662D0BC21D0261AE3B6`
and `K1A_USB01.csv` is
`E784193F5C0A1D9084E5D11543D9F1758F15B1784CB35EAF4E9E918E202ADDDE`.
All PID captures and earlier CSVs remain readable. No `.tmp`, `.part`, or
`.bak` artifact exists on either drive.

## Disposition

32 KiB Round 2 passes end to end on the opposite hardware pairing: WAV
load/playback, PID, full image replacement, full-range 32 KiB HPT
publication, filesystem consistency, exact content, internal image CRCs, and
independent removal/reinsertion validation all pass.

Controller A's initial below-class `USB05_4G` enumeration miss and the single
clean powered reseat remain part of the result; they are not erased by the
successful FileX and Pi gates. No VBUS, controller power, format, repair,
optional-SCSI, firmware retry, or drive-specific code path was used. Rounds 1
and 2 now supply four of the required eight 32 KiB controller/drive
observations. Broad promotion remains blocked on Rounds 3 and 4 and final
disposition of the repeated Controller A first-insertion enumeration pattern.
