# USB BOT representative delta regression

Status: complete
Result: complete; current broad USB production baseline promoted
Type: evidence
Updated: 2026-08-01
Next Action: retain this baseline; reopen architecture review for any newer contradictory standards-conforming-device incident
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

## Result

PASS. The isolated active-path BOT CSW validation candidate passed its
representative cross-device delta regression without a reseat, enumeration
retry, VBUS action, remount sequence, optional SCSI lifecycle, media write, or
specimen-specific production behavior.

This promotes the candidate as the current broad USB production baseline. The
claim is bounded: standards-based command and recovery policy plus the
representative evidence pass; it is not a claim that every possible USB drive
or every future failure is impossible. A newer contradictory incident from a
standards-conforming device reopens architecture review rather than creating a
drive exception.

## Frozen candidate and deployment

- Source commit: `ca6f573a`
- Release ELF SHA-256:
  `68AC029578BE8EDB7CD3F5B39254C8A4B5351A5E9161E87DDE7A1D2DC6ECF0A9`
- Release S-record SHA-256:
  `6907F88FB91D5DC58EBF15A918C3993FDE3A076AFAE9DAEFB099899F55932581`
- Controller A J-Link `69664082`, hardware variant `0`: serial-bound
  programming and independent full 4 MiB readback passed.
- Controller B J-Link `69660123`, hardware variant `1`: retained the same
  candidate and independent full 4 MiB readback.
- Both full-flash SHA-256 values:
  `23428A890890CA40172483A683F2B985C5BFC640650E7B33B977EAD15DB754CA`.

Predeployment, postdeployment, and final captures found the correct hardware
variant, HPT-ready state, zero application/system errors, and no USB attached
where required.

## Representative observations

| Observation | First insertion | Application read | Final attached state | Pi closeout |
| --- | --- | --- | --- | --- |
| `USB03_32G` / Controller B | PASS, configured high speed, acceptance `15`, recovery `0` | WAV load/playback accepted | mounted, valid session, zero errors | unmounted, read-only FAT check, equal FATs, clean kernel log |
| `USB03_32G` / Controller A | PASS, configured high speed, acceptance `15`, recovery `0` | WAV load accepted | mounted, valid session, zero errors | unmounted, read-only FAT check, equal FATs, clean kernel log |
| `USB01_2G` / Controller B | PASS, configured high speed, acceptance `15`, recovery `0` | WAV load accepted | mounted, valid session, zero errors | unmounted, unchanged equal FATs, clean kernel log |
| `USB05_4G` / Controller B | PASS, configured high speed, acceptance `15`, recovery `0` | WAV load accepted | mounted, valid session, zero errors | unmounted, unchanged equal FATs, clean kernel log |
| `USB02_8G` / Controller B | PASS, configured high speed, acceptance `15`, recovery `0` | WAV load accepted | mounted, valid session, zero errors | unmounted, unchanged equal FATs, clean kernel log |

Every controller observation enumerated `W00.WAV`, `W01.WAV`, and
`7520553_SMI_00_05_00.imx`. No first insertion failed below class and no
recovery attempt was needed. This delta intentionally reused the prior full
A/B write/publication matrix rather than rewriting the media solely to qualify
the BOT validation change.

## Independent media invariants

The Pi identified each drive by canonical label plus its established model,
serial where available, capacity, partition identity, and FAT UUID. All checks
were unmounted and non-correcting. `fsck.fat -n -v` reached the final
filesystem sector and explicitly left each volume unchanged. Exit `1`, where
present, was only the established Windows root-directory label versus blank
boot-sector-label convention.

| Drive | FAT bytes per copy | Entry and exit SHA-256 |
| --- | ---: | --- |
| `USB03_32G` | 7,574,016 | `A03BB7CA83EFDBA71BF7120A5C9078E48A00442334935E4EE0AB434C0CE5B725` |
| `USB01_2G` | 1,961,472 | `B6A43EB2B907E1E09788026A4505C71EB1A01D1E5AED2F400829665A6DC0C312` |
| `USB05_4G` | 3,923,456 | `DC204E88B1F985E62BBB124DE59341B0D83C9D6489119B27B86117543EDC7476` |
| `USB02_8G` | 7,847,936 | `07B68DB6930553CBE166B674227602375F42192865B21F9578D199EA1EA92E7F` |

Each pair of FAT copies was byte-identical at both the entry and exit gates.
Current-insertion Pi logs contained normal enumeration/capacity/cache events
and no USB reset, timeout, medium, offline, descriptor, sense, buffer-I/O, or
block-I/O error.

## SD identity and final controller state

Controller A final baseline:

- protocol card status `5`, CID-valid `1`, representative-sector-read `1`;
- MID `0x03` (SanDisk), OEM/application ID `SD`, product `SS08G`, revision
  `8.0`, serial `2540634244`, manufacturing date `2024-05`;
- raw CID `035344535330384780976F0084018559`.

Controller B final baseline:

- protocol card status `5`, CID-valid `1`, representative-sector-read `1`;
- MID `0x03` (SanDisk), OEM/application ID `SD`, product `SS08G`, revision
  `8.0`, serial `2540634245`, manufacturing date `2024-05`;
- raw CID `035344535330384780976F0085018507`.

Both final captures had no USB attached, cleared FileX ownership, HPT status
`0`, zero USBX/application/system errors, and zero enumeration-recovery
attempts. Raw target captures remain under the ignored paths
`work/evidence/hpt-log-store-circular-mirror/controller-a/usb-bot-representative-delta-20260801/`
and
`work/evidence/hpt-log-store-circular-mirror/controller-b/usb-bot-representative-delta-20260801/`.

## Disposition

The current production baseline combines:

- the minimal common-denominator production lifecycle;
- the proven 32 KiB FileX HPT write cap;
- recoverable application-session handoff;
- one bounded, once-per-attachment enumeration retry for the proven terminal
  below-class failure;
- strict BOT CSW validation and USB-IF reset-recovery order; and
- fail-closed essential-command, transport, FileX, and content handling.

The established full USB09 and four-drive/controller write/publication matrix,
deterministic enumeration proof, deterministic malformed-CSW coverage, and
this cross-device delta contain no unresolved newer contradictory incident.
The broad USB compatibility recovery workpackage can therefore close.
