Skip to content

P4_CLOSEOUT.md

Latest commit

 

History

History
112 lines (73 loc) · 4.46 KB

P4_CLOSEOUT.md

File metadata and controls

112 lines (73 loc) · 4.46 KB
Raw

ExecSurface — P4 Backend Adapter / Authority Architecture Closeout

Date: 2026-09-30 Parent program: #100 Branch: development/post-alpha4-behavioral-integrity Protocol: docs/development/P4_BACKEND_AUTHORITY_PROTOCOL.md

Status

SCIENTIFICALLY CLOSED — BOUNDED RESEARCH-ONLY

P4 closes without public integration or release promotion.

Protocol-conformant outcomes

The evidence supports the following P4 outcomes within their stated scope:

  • P4_PTRACE_ADAPTER_CONTRACT_ESTABLISHED_BOUNDED
  • P4_EXTERNAL_IMPORT_PATH_ESTABLISHED_BOUNDED
  • P4_BACKEND_EQUIVALENCE_NOT_ESTABLISHED

For the three A2 success-authority gaps actually tested, the evidence also supports the narrower operational conclusion that no second backend is justified by those gaps. This is not a claim that no future proposition can ever justify another backend.

P4_PROPOSITION_SCOPED_MULTI_BACKEND_PATH_ESTABLISHED is not claimed: no second collection backend received success authority for the tested propositions.

What P4 established

A — authority architecture

P4 separated semantic authority from collection-backend identity:

  • proposition-scoped evidence rather than bare event presence;
  • explicit authority and completeness;
  • deterministic evidence/proof identity;
  • unsupported, ambiguous, pending, failed and lost evidence remain distinct from success;
  • backend/profile names cannot upgrade authority;
  • representation gaps do not automatically justify another collector.

B — ptrace success-authority requalification

The three A2 gaps were requalified under bounded research-only ptrace experiments:

  1. P4.FILE.OPEN_OBJECT
  2. P4.FILE.RENAME_DELETE
  3. P4.NET.CONNECT_DESTINATION

B1, B2 and B3 each passed their mandatory live Linux x86_64 corpus and model/falsification gates. Cross-proposition falsification then passed 12/12, including replay, substitution, FD reuse, proposition confusion, attempt/pending/failure laundering, observer loss, causal substitution and deterministic identity attacks.

Bounded P4-B decision:

P4_B_PTRACE_SUCCESS_AUTHORITY_ESTABLISHED_BOUNDED

Boundary:

NO_SECOND_BACKEND_JUSTIFIED_BY_THE_TESTED_P4_B_GAPS

C — external trace interoperability

P4-C demonstrated a schema-level import path from one pinned mature external source without treating producer identity or event presence as global authority.

Accepted P4-C evidence:

  • source SHA: 7db008ad64aad0782b7438ea8f225739ab942fc5
  • run: 36739233309
  • job: 109968900093
  • artifact: 11110450417
  • artifact digest: sha256:3aaaf91a7ea3f6dc28636bd864be7a227b3129545a39d885f305f30f7bcda043
  • frozen import corpus: 14/14 PASS
  • P4-B cross reproof: 12/12 PASS
  • B1/B2/B3 live reproof: 10/10 / 10/10 / 10/10 PASS
  • Semantics v3: 7/7 PASS
  • public M11: 6/6 PASS

Decision:

P4_C_EXTERNAL_IMPORT_PATH_ESTABLISHED_BOUNDED

The external-import result remains incomplete at session/transport level unless stronger evidence proves those dimensions. It does not establish Tetragon/ptrace equivalence or authorize Tetragon as a required backend.

Negative evidence retained

P4 closeout does not erase failed runs or counterexamples. Retained evidence includes, among other items:

  • P1/eBPF negative/incomplete evidence that prevented unjustified backend promotion;
  • B3 zombie-process lint failure;
  • B3 loopback ETIMEDOUT 9/10 run;
  • B3 test-adequacy defect and fixture byte-order correction;
  • P4-B rustfmt-only failure;
  • P4-C rustfmt-only failure;
  • P4-C large_enum_variant clippy failure;
  • P4-C 13/14 test-fixture failure where the mismatch fixture accidentally mutated both identities.

Each correction remained bounded to the demonstrated defect and did not weaken acceptance criteria.

Immutable/public boundary reverified

P4 does not modify or promote:

  • public release v0.1.0-alpha.4;
  • immutable release source 48e0b9a0707553349e75e97a9dfa096d13f9ab5d;
  • stable Action AETHERXGLOBAL/execsurface@v0.1;
  • public default observer;
  • raw/canonical/baseline v2 meaning;
  • historical evidence.

No merge to main is authorized by this closeout.

Anti-drift conclusion

P4's useful result is not “more collectors.” It is a bounded authority architecture in which evidence sources remain interchangeable only where proposition-level proof exists.

The next scientific phase is P5 Runtime Attestation / Provenance Interoperability. P5 must reuse existing ecosystem standards wherever they fit and must not invent a new predicate merely to create project-specific surface area.