# Controller A HPT2 S5 runtime matrix pass — 2026-08-05

Status: pass; Controller A S5 production-queue contention minimum is 10/10
Type: evidence
Updated: 2026-08-05
Next Action: collect Controller A S4 checkpoint-region and S6 slow-path evidence; keep Controller B disabled until every A-local gate passes
Campaign Reason: Historical record from campaign 009-complete-clean-contract-bound-a-b-qualification; retained independently after campaign closure and requires no separate owner-queue execution.
Campaign: standalone
Parent: work/checklists/checklist-hpt2-bounded-stop-recovery-and-regression-measurement.md
Canonical Truth: docs/hpt2-measurement-matrix-v1.md

## Result

Controller A completed ten separately admitted S5 observations under exact Pi
release `hpt2-runtime-map-6dc1c3c` and unchanged firmware bundle
`smi-00-05-00-52edddc39e30-054d5386ad9d-release`. Every observation proved two
production logs queued and two completed, zero queue drops, zero HPT store
loss, an empty queue at both boundaries, clean terminal reason `0`, and terminal
proof `0x1F`. The S5 gate is `10/10` and passes.

Each run contains exactly one acknowledged Start and one acknowledged Stop
operation. Ambiguous post-write status captures were resolved only through the
stored operation's read-only reconciliation path; no command was replayed. All
ten runs preserve CID `05000C61736466672200001C8901A34F` and boot-session
witness `DEE10A12EC443248`.

The S5 Stop-to-terminal observation upper bounds range from `2020 ms` to
`2054 ms`. The ordinary boundary adjudicator separately records nine S2 shapes
and one S1 shape; it does not infer S5 from the planned label. Only the extended
controller-runtime adjudications below grant S5 credit.

## Compact immutable manifest

| Run | Extended evidence | SHA-256 | Upper bound | Base shape |
| --- | ---: | --- | ---: | --- |
| `0703f83e-16a9-4ea2-9022-f337a5937f00` | 672 | `44e4e8eefd79d42021497132d48e301d97f0f5bb624413f591aeb31f1914096b` | 2023 ms | S2 |
| `ec44a69c-46a9-4394-99e5-87dcc273db11` | 689 | `3bc3ad943fb0996686b92c713e76d115b9df8545d6e2483822ca28a6de388d1a` | 2036 ms | S2 |
| `db0b3bdc-f2fe-4a6a-9261-546e7046203f` | 706 | `bfedb78190cd4d42b1eea463b66d1645d4ee4024e74a22866d71f189f6081887` | 2035 ms | S2 |
| `ef4ecebf-9486-4fba-9420-a0eb499eb9e2` | 723 | `c7d573d9d4c2c10781adef73cf3765b3ef1c5ace3b4361daa6d5a02cf6e38c70` | 2024 ms | S2 |
| `5ff2f46b-63e5-4415-bd9b-0c0d835ec5c9` | 740 | `4b283fc30edee129630ee7495bd2c857965aa6f60486793322cba82ec8f5a177` | 2020 ms | S2 |
| `405fb096-25a7-4395-bc1b-05dc827f086c` | 757 | `4e4eebb2a0c42c7dca5a94efa715098ec33b793cf24d74c0bafe497e9c11b50b` | 2029 ms | S2 |
| `40ece649-bc4c-46d1-9301-67458d9923d0` | 774 | `bd71eab6487ef96ddb412e0086db46fa736665e625405da74a9a1659a634d0c7` | 2024 ms | S2 |
| `df78c7ba-7b14-49d8-805d-c4340cb04732` | 791 | `87dc0d06afbe3fa983155c91fcd581097591226b233f6faa0e2ca25928e021e0` | 2022 ms | S2 |
| `afc90c0a-2557-4e9f-a34c-708ad114b95a` | 808 | `66a82252b952654825989af8caa36ac50b2cabeefe71a5cdf86227dafbf101a5` | 2023 ms | S2 |
| `4a868d99-f4db-4b7f-86e9-38e759feeed4` | 825 | `438fce8cd760645da60c2b7ed33cf953e7f5396924548862e6092c07c5322db7` | 2054 ms | S1 |

Every row records queued/completed/drop/loss deltas `2/2/0/0`. Each extended
artifact binds its own immutable baseline and terminal runtime artifacts, exact
map SHA-256
`ca774f3fd897487ab41161b161bdecd83fffc3146db9347074f815e1620b0dc4`,
firmware bundle, controller role, CID, and boot session.

## Live audit and cleanup

The live audit now reports 45 base adjudication rows and ten valid extended
rows. Controller A is S0 `3/3`, S1 `13/10`, S2 `20/10`, S3 `12/10`, and S5
`10/10`; S4 and S6 remain zero. The A-local gate therefore remains false,
Controller B execution remains disabled, the overall matrix remains incomplete,
and `stop_deadline_ms` remains `null`.

Postflight found zero active runs, no controller tool container, and the default
non-loopback firewall drop intact for ports `19120` and `19121`. Controller B
was not accessed.
