# G-PWR-B full reset-path qualification result

Status: pass; G-PWR-B gate removed
Type: evidence
Updated: 2026-07-25
Next Action: use this qualified power method only for otherwise authorized post-freeze cases
Campaign Reason: Historical record from campaign 013-close-hpt-media-2-initial-release-qualification; retained independently after campaign closure and requires no separate owner-queue execution.
Campaign: standalone
Parent: work/runbooks/runbook-autonomous-hpt-sd-qualification.md

## Result

Controller B / Setup B passed three consecutive, independent coordinated
power cycles on the current candidate firmware. Every cycle removed the exact
J-Link, observed the required 10-second and 30-second intervals, restored Plug
2 and J-Link power, reconnected to the exact target over SWD, cleared the
volatile reset witness, retained the expected protocol/media state, and
returned the same complete SD-card CID.

`G-PWR-B` is therefore closed as a qualified power/reset method. It is no
longer an unverified or minimal-equivalence gate.

## Candidate and target identity

- Controller assembly: `1751219`
- J-Link serial: `69660123`
- Source commit recorded by the harness:
  `4896fe86170b6e2128c80843d18e1f8a2ba38b9f`
- Debug S-record SHA-256:
  `1D047B144B6DFA88447212ED2670C3F7B636D24D29B5A20B100453A4806B2978`

The first programming command exceeded its host-side timeout. The exact same
image was then retried with a longer bound; SEGGER reported that both option
memory and program flash already matched, then reset and started the target.
No alternate image was loaded.

## SD-card identity

The baseline and every post-reboot capture agree:

- Protocol card status: `3`
- CID-valid status: `1`
- Representative-sector-read status: `0`
- Manufacturer: SanDisk/WD (`MID 0x03`)
- OEM/application ID: `SD`
- Product: `SS08G`
- Product revision: `8.0`
- Serial number: `2540634245`
- Manufacturing date: `2024-05`
- Raw 128-bit CID: `035344535330384780976F0085018507`

The baseline record is `baseline/sd-card-identity.json`. The three corresponding
post-reboot records are under each `witness-N/post-reboot/` directory.

## Consecutive reset witnesses

| Witness | Coordinated operation | J-Link off | Plug off | Plug on | Target restored | Result |
| --- | --- | --- | --- | --- | --- | --- |
| 1 | `bd2029ae-9c3a-4084-bdbc-7d8f2a0df85c` | 12:06:36Z | 12:06:46Z | 12:07:16Z | 12:07:32Z | pass |
| 2 | `3c174e88-8b79-4a18-a648-31d4b6c303c2` | 12:08:55Z | 12:09:05Z | 12:09:35Z | 12:09:52Z | pass |
| 3 | `3204fcc1-2bd5-469c-8224-e37075758505` | 12:13:34Z | 12:13:45Z | 12:14:15Z | 12:14:32Z | pass |

For all three witnesses:

- post-cycle command and status were both `0`, proving startup cleared the
  deliberately armed RAM witness;
- SD protocol state returned to expected state `3`;
- HPT media state/status returned to their expected values `1/5`;
- Plug 2 ended on;
- J-Link `69660123` ended powered and present; and
- bounded SWD connection to `R7FS7G27H` at 4 MHz passed.

## Scope boundary

Controller B's installed card remains FAT/FileX media, reported as HPT media
state/status `1/5`. This qualification did not format or initialize the card,
because destructive media authorization was not given. It therefore qualifies
the Group B power/reset path and current-firmware reboot behavior, but does not
claim Controller B HPT-media initialization, mirrored-data durability, or
endurance qualification.

The storage-boundary correlation supporting the shared G-PWR method is the
Controller A identical-firmware data-durable-before-commit result in
`controller-a/g-pwr-20260725/qualification-result.md`. Controller B HPT
media/data qualification remains a post-freeze card-specific activity, not a
remaining G-PWR gate.
