P3-H P0.1: encoder-produced deterministic SV WirePlan

[masarray]

Stacked on #140. Tracks #139.

Base head expected: a09e0312dd1257404669ddc2e6a2cf4264d072a6
This PR must not merge before #140 is accepted; after #140 lands, retarget/re-evaluate against current main.

Why

The deterministic ESP32-P4 publisher still rediscovered smpCnt, confRev and seqData offsets by scanning the fully encoded frame for tag/length byte patterns. Separately, the smpSynch TX interception shim structurally parsed BER on every transmitted SV frame and contained a second canonical transport-failure guard even though app_main.cpp already owns that failure policy.

That is correctable technical debt in the realtime path.

Architecture change

• SampledValuesPduCodec::encode_into_with_layout()
  • records patchable ASDU value regions at the exact moment BER bytes are written;
  • caller supplies std::span<SampledValueAsduEncodeLayout>; no metadata heap allocation;
  • layout is naturally per-ASDU and therefore multi-ASDU-ready.
• SampledValuesFrameCodec::encode_into_with_layout()
  • preserves the same encoded bytes;
  • translates PDU-relative regions into absolute Ethernet-frame offsets.
• ESP template build consumes this metadata once while STOPPED.
• Realtime publisher patches smpCnt, smpSynch and seqData directly by bounded offsets.
• The compile-time TX interception macro and per-transmit BER scan are removed.
• The duplicate transport-failure guard in smp_synch_lab.cpp is removed; app_main.cpp remains the single canonical TX failure authority.
• The existing structured wire_field_offsets.hpp utility remains available for inspection of arbitrary/foreign frames; it is no longer part of deterministic publication.

Runtime properties

No XML/SCL parsing, BER parsing, packet rescanning, dynamic re-encoding or new allocation is added to the steady-state publish loop.

Per published sample, synchronization policy resolves from atomics and the selected value is a direct one-byte patch. Canonical/mirror frames use the same decision for that sample.

Regression coverage

• layout encoding produces byte-for-byte identical output to existing encode paths;
• frame-relative offsets are validated for VLAN and untagged frames;
• field presence/absence is validated for optional ASDU fields;
• two-ASDU layout is validated and both smpCnt fields can be patched independently then decoded correctly;
• insufficient caller layout storage fails closed;
• current bounded firmware still requires exact widths: smpCnt=2, confRev=4, smpSynch=1, seqData=64.

Non-goals

No IEC 61869-9 profile expansion, no arbitrary dataset activation, no multi-stream scheduler, no PTP conformance claim and no PRP/HSR work in this PR.

Next node after P0/P0.1: #139 P1 timing/profile semantic separation.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-06T12:48:13.281940Z	d63cec7	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.