Nothing lives outside the repo: the bench ixgbe driver source moves in tracked at kernel/ixgbe-sff/ (build artifacts gitignored, load-ixgbe repointed, module build verified from the new home — the reg_ops/ts_bench bench extras are finally versioned), the ice sff_i2c patch is vendored at docs/nics/e810/ice-sff-i2c.patch with its out-of-tree build recipe (TRACE_INCLUDE_PATH fixup, version-matched Makefile), the e810 NVM flash logistics land in the e810 README (exact package name, rollback -m B49691D578EC into the inactive bank, post-flash inventory failing until power cycle is the expected tell), and every ~/work/phydiag-work pointer becomes an inline statement; phydiag-work, ice-sff, e810nvm and the wiitekphy CLI (a duplicate of documented transport+protocol) are deleted

docs/modules/fs/README.md

@@ -77,7 +77,7 @@ Protocol and full verified catalog: [bcm84891l-mdio-commands.md](bcm84891l-mdio-
 
 **The internal poll's clock, measured on hardware.** One poll every 3.49–4.22 s, regime-dependent but rock-stable within a regime (±15 ms): ≈3.50 s warm and idle, ≈3.70 s, ≈4.01 s cool and idle, ≈4.21 s under continuous host mailbox load — host traffic *defers* the poll, never advances it, and nothing observed brings two polls closer than 3.49 s. The poll is link-independent: it runs straight through an AN restart and the whole retrain without missing a beat. Around each poll's service the bridge serves stale data for ~50 ms: every read returns the last value successfully fetched before the µC went busy (stuck-at-last-fetch, not one-behind), which is exactly how 3.33 inherits 1.147's 0x0011. The window closes *before* the poll is detectable — a CMD read returning the fresh resident 0x0031 is itself proof the µC serviced the fetch, so flip detection doubles as the all-clear.
 
-**The windowed protocol (`phy.go`) — contention resolved by time-division.** Since the poll cannot be silenced, the host schedules around it. One goroutine per module owns every transport touch (requests execute one at a time on the loop — nothing else can reach the wire, by construction), and each request is admitted only inside a fixed 3.4 s window following an observed poll (CMD flipping to resident 0x0031, polled at 10 ms); at the cutoff the host goes quiet until the next flip. 3.4 s sits below every observed cadence, so a regime switch mid-run lands harmlessly in vacated time — no prediction, no period tracking, just the last observed flip plus two constants. **Re-lock — no window at assumed phase**: edge detection needs CMD ≠ 0x0031, and handler commands leave it armed (every one rewrites CMD); whenever CMD sits at the resident 0x0031 at expiry the phase is unknown and admission arms first, then anchors on the true edge that follows. Steady-state polling issues no handler commands of its own (the SNR GET is dropped), so each taken edge immediately re-arms with a bare GET_PAIR_SWAP: the single CMD write per window lands at the start of the quiet period, maximally far from the next poll — the wedge risk is writes near the poll — and every later expiry reads the phase without writing. One command per ~4 s window replaces a handler GET per second. Admission then arms first (a bare GET_PAIR_SWAP, resident 0x0000) and anchors only on the true edge that follows, so every window is edge-locked from first contact at boot; bringup pays one phase acquisition per module (~2–4 s). A missing heartbeat free-runs one window rather than stall (never observed — the heartbeat survives retrains). Before re-lock the post-diag window anchored blind on the stale resident, and every ghost observed under windowing sat in exactly that window; with re-lock, ghost-free across all bench diags to date. A poisoned batch is self-consistent and forensically visible: the stuck value clears the link bit (SNR skipped, panel shows "-", the ghost-SNR panic sidestepped) while 3.33 charges the same value as errored blocks (0x0011 → the "+17"); every corrected charge and link-down reading prints its raw 1.1/3.33/1.147 on the console, so events self-attribute — a ghost names its neighbor register, real correction decodes as nothing but itself. Validated on hardware in `~/work/phydiag-work/bcm_phaselock_bench.py`: aiming read bursts at the poll reproduces the poisoning on 100% of polls (the "+17" manufactured on demand); windowed operation ran 2,881 back-to-back batches — 16× the production rate, riding the cutoff — with zero stale values, and the production 1 Hz stream is unperturbed (blackout hides inside natural gaps, worst sample gap ~1.6 s).
+**The windowed protocol (`phy.go`) — contention resolved by time-division.** Since the poll cannot be silenced, the host schedules around it. One goroutine per module owns every transport touch (requests execute one at a time on the loop — nothing else can reach the wire, by construction), and each request is admitted only inside a fixed 3.4 s window following an observed poll (CMD flipping to resident 0x0031, polled at 10 ms); at the cutoff the host goes quiet until the next flip. 3.4 s sits below every observed cadence, so a regime switch mid-run lands harmlessly in vacated time — no prediction, no period tracking, just the last observed flip plus two constants. **Re-lock — no window at assumed phase**: edge detection needs CMD ≠ 0x0031, and handler commands leave it armed (every one rewrites CMD); whenever CMD sits at the resident 0x0031 at expiry the phase is unknown and admission arms first, then anchors on the true edge that follows. Steady-state polling issues no handler commands of its own (the SNR GET is dropped), so each taken edge immediately re-arms with a bare GET_PAIR_SWAP: the single CMD write per window lands at the start of the quiet period, maximally far from the next poll — the wedge risk is writes near the poll — and every later expiry reads the phase without writing. One command per ~4 s window replaces a handler GET per second. Admission then arms first (a bare GET_PAIR_SWAP, resident 0x0000) and anchors only on the true edge that follows, so every window is edge-locked from first contact at boot; bringup pays one phase acquisition per module (~2–4 s). A missing heartbeat free-runs one window rather than stall (never observed — the heartbeat survives retrains). Before re-lock the post-diag window anchored blind on the stale resident, and every ghost observed under windowing sat in exactly that window; with re-lock, ghost-free across all bench diags to date. A poisoned batch is self-consistent and forensically visible: the stuck value clears the link bit (SNR skipped, panel shows "-", the ghost-SNR panic sidestepped) while 3.33 charges the same value as errored blocks (0x0011 → the "+17"); every corrected charge and link-down reading prints its raw 1.1/3.33/1.147 on the console, so events self-attribute — a ghost names its neighbor register, real correction decodes as nothing but itself. Validated on hardware: aiming read bursts at the poll reproduces the poisoning on 100% of polls (the "+17" manufactured on demand); windowed operation ran 2,881 back-to-back batches — 16× the production rate, riding the cutoff — with zero stale values, and the production 1 Hz stream is unperturbed (blackout hides inside natural gaps, worst sample gap ~1.6 s).
 
 ## ECD — recovered from the OpenBCM SDK, proven on hardware
 
@@ -96,7 +96,6 @@ Observed on a plugged, linked, healthy cable: verdicts OK×4 and per-pair length
 - **An interrupted run can wedge the µC's SMI service** (observed after a process died mid-diag): the bridge ACKs I2C but serves 0x0000 for every register at any delay and silently drops MDIO writes — no in-band recovery, not even PMA reset — while the EEPROM path and the trained link keep working. Recovery: a driver unbind/rebind of the port (the SFP re-initialization resets the µC); a reseat would do the same.
 - Fault verdicts (open/short/inter-pair) are unexercised — deliberately: the product is a closed-loop tester, both ends always plugged.
 - Family constraints from the SDK: port must be enabled; unsupported at forced 100M.
-- `bcm_ecd_probe.py` in phydiag-work implements the recipe.
 
 ## Missing datasheet chapters
 

docs/modules/wiitek/README.md

@@ -32,7 +32,7 @@ The shared IEEE 802.3an set ([../README.md](../README.md)) plus two safe vendor
 
 Proven-safe writes, all IEEE standard space: AN restart `7.0 |= 0x1200`, and PMA reset `1.0 = 0x8000` — **the PMA reset clears the speed-selection bits** (1.0 reads 0x0000 after, no speed selected, 10G training impossible) and does not touch 7.x config; write 1.0 = 0x2040 back after any reset or the link never returns.
 
-`rb_forensic.py` in phydiag-work performs this baseline and logs every sff_i2c op (PRE line fsync'd before each op, POST after) to `wiitek-ops.log` — a PRE with no POST names whatever killed the µC. `rollball_ixgbe.py` hard-guards the `*.0x??64` blacklist and raises before touching hardware.
+The forensic method for this baseline: log every sff_i2c op with a PRE line fsync'd before the op and a POST line after — a PRE with no POST names whatever killed the µC. Any client must hard-guard the `*.0x??64` blacklist and raise before touching hardware.
 
 ## In the test set — mixed pair with the FS
 
@@ -42,7 +42,7 @@ Not available in safe space: pre-FEC counter, cable length — those live in the
 
 **A down port hides the PHY: TX_DISABLE powers the core down, not the µC.** ixgbe qualifies this module as fiber (the fake-SR EEPROM) and asserts TX_DISABLE while the netdev is admin-down; the module then answers every PHY register read with 0xFFFF through a fully working, echo-matched mailbox — EEPROM reads are unaffected. Proven by flipping admin state on the bench: down = PHY ID 0xFFFFFFFF, and the core wakes within 0.25 s of the raise (measured 0.07–0.24 s across cycles; the BCM answers instantly). The dev host masks it (its unmanaged config leaves the ports up from boot); a fresh appliance boot hits it on a never-raised port. cabletest therefore raises the links before first touching a module, and settles identify through a short bounded wait while the core wakes.
 
-**Mailbox latency is regime-dependent** — <25 ms per read linked (X520 host), ~220 ms steady with no link (bench-measured, `~/work/phydiag-work/rb_mbox_latency.py`), and beyond 3 s while 10GBASE-T training runs (observed live when a cable plug landed mid-read at bringup). cabletest's mailbox deadline (`rbCmdTimeout`) is sized past a full ~6 s training so a plug event reads busy, not dead.
+**Mailbox latency is regime-dependent** — <25 ms per read linked (X520 host), ~220 ms steady with no link (bench-measured), and beyond 3 s while 10GBASE-T training runs (observed live when a cable plug landed mid-read at bringup). cabletest's mailbox deadline (`rbCmdTimeout`) is sized past a full ~6 s training so a plug event reads busy, not dead.
 
 **Mailbox trap: an orphaned command's completion answers as yours.** If a session dies mid-mailbox-op (killed process), the µC finishes the orphaned command on its own time, leaving CMD=DONE and that command's value in the block. A next session that issues a command and samples status immediately reads the orphan's DONE and value as its own — seen live: a PHY ID 1.2 read answered 0x0006, the orphaned EEE-advert (7.60) value of the killed session, while the very next read was correct. The client therefore never samples status before a full poll gap, reads the value from the same 6-byte block read as the status, and accepts a completion only when the block echoes the issued devad/reg (`phy.go` mbox). The split-transaction python clients always had the pre-sample gap, which is why the field campaign never hit this.