Worker B: 26 new this cycle, incl. the rare-epilogue three-way model
confirmed textually (sub-case b: 0x800F8928 mutual exclusion, before/after
68->64 then or-encoded moves), the discard-an-argument-outright pass-through
5th flavour, and D_80121ED8 corroboration. Worker C: 4 matched (0x80048E20,
0x8005784C, 0x800579A0, 0x80057AE0). GATE INCIDENT: C's 0x8002E3FC claimed
but its source is LENGTH-MISMATCH (88 vs expected) against the tracked
registry — dropped from the merge, reported back to C. Gate MATCH
whole-binary SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
4 tier-2 frames (0x8002864C, 0x80049298 with gp=-D_80121BFC, 0x801032D4,
0x800893E8). C's process-error owning (hardcoded ends instead of the
worklist column) and the F24 finding (caller-saved reloaded pointer across
a call means the store belongs before the call) recorded. Rare-epilogue
census: 47 tier-2 rows carry the pattern at their final instructions —
definitive class, broadcast to the workers. Gate MATCH whole-binary
SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker C closed 0x80031F2C (a row the coordinator had recorded as a
negative) via the F23 struct-assignment lever (second confirmation) —
stale negative removed. Also 0x800129C8, 0x8009132C (F22: unsigned char
local re-masks an already-zero-extended lbu — width-of-DESTINATION tell).
C's link-error finding recorded: 'small-data section too large' on a
region that built fine = a duplicate gp symbol (name mismatch), not a -G
problem. New negative 0x80042E68 pairs with 0x80042DD4; named pointer
lever untried. Gate MATCH whole-binary SHA-1
e173426c157384ebf1b6caf8c6fea18a85a14af9.
3 tier-2 frames from C (0x800A6840, 0x8002C728, 0x8002E968). Gate DIFF
avoided: C's 0x8002C728 used the name g_80121B84 which resolves implicitly
WITHOUT the gp marker (F19) — its local superset carried a g_-marked row so
C's own run matched, but the tracked registry only has D_80121B84. Fixed the
source to the tracked name; re-verified MATCH. C's 0x80025070 'negative'
corrected: that row is the coordinator's registered claim (stale index
entry removed). C's address-arithmetic charter warning (lui + negative
displacement must be hand-derived; five occurrences) recorded. Gate MATCH
whole-binary SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker B: 3 more tier-2 frames (0x80036328, 0x800AA2B4, 0x80021FF4 — the
latter with the base+offset decode and the clearest finding-13 scratch
constant yet) + symbol D_80121978 (gp). 23 new bodies this cycle.
GATE INCIDENT: the first candidate gate DIFFed with 4 extra bytes because
src/func_80025070.c was DELETED from the working tree after my commit
( status) — the build used a stale/wrong object. Restored from git and
re-gated clean. Lesson recorded: every tracked src/ file is a claim; a
registry row whose source is missing breaks the whole-binary gate with a
cascade, so the pre-gate 'git status src/' check must catch deletions too.
Gate MATCH whole-binary SHA-1
e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker B's maspsx structural limitation recorded: a region needing
addu-encoded moves (maspsx expands move->addu, GNU as expands move->or)
AND an assembler-filled jr slot cannot be matched with any single
maspsx= setting — this explains a family of epilogue residuals. B
verified in .s files. Two negatives with mechanisms (0x800254B0 alloc,
0x800751E8 with negu — evidence FOR the toolchain, finding-26 route).
B self-corrected an invented-address method error. Gate MATCH whole-
binary SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker B matched 0x800F6330 (correcting my stale negative: destination is
0x8011AA18 via the signed-displacement finding-4 trap) and 0x80090B7C
(18th/19th first-attempt matches this cycle). Coordinator absorbed
0x80025070 (conditional dispatch wrapper). B's correction of its own
void-vs-valued epilogue tell accepted (0x800FBDC0 counterexample; the
shared-epilogue hypothesis stands untested — recorded as observation, not
rule). Gate MATCH whole-binary SHA-1
e173426c157384ebf1b6caf8c6fea18a85a14af9.
Two more coordinator tier-2 wrappers, both first-try: 0x800A623C
(symbol-form constant + pass-through swap), 0x800A5CEC (guarded call with
zero in the jal slot). Gate MATCH whole-binary SHA-1
e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker B: 0x8002E3A8 bare unused frame (16B of pure stack adjustment from a
removed local — reproduced by declaring char unused[16] and nothing else),
0x8010AAC0 pass-through with arg reorder, 0x8008FF58 six-arg rebuild,
0x800F79C0 virtual dispatch. Worker C: 0x800268C4 (D_ names), 0x800697C4,
0x800658FC (F20 displacement-fold lever), 0x800A5CC8 (overlap with the
coordinator's claim — C's registration stood). Gate MATCH whole-binary
SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
B's alignment-1 find closed 0x80013D04 first-attempt (lwl/lwr on provably
aligned addresses = declared-type alignment, not real alignment; Quad4
struct model). New symbol D_801219F4 (gp-relative address taken without
lui). Coordinator wrappers 0x800F8FC0/0x800A5CC8 (pass-through + constant-arg
delay-slot). Report negatives reconciled into the tracked index (0x800F9134,
0x80094370 now excluded by regenerated worklist). Gate MATCH whole-binary
SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker B observed that gating each batch without regenerating the worklist
re-grows partition drift every cycle (their file was 10 rows stale). The
regeneration is now done alongside each merge; partitions re-proved disjoint
(486/486/486, total 1458).
4 new (0x80082750, 0x8007EB8C, 0x800B34FC, 0x800C1EA8), 2 new gp symbols
(g_80122610, g_80122618) for the gp-relative table-address rows. Recorded
findings F17 (sra-vs-srl = signedness of the shifted value), F18
(inline mask strength-reduce -> bind to a local), F19 (address-shaped name
resolves implicitly but loses its gp marker -> use the registry's exact
name). Gate MATCH whole-binary SHA-1
e173426c157384ebf1b6caf8c6fea18a85a14af9.
Per-guard inline reload of obj->p (finding-8 CSE prevention) + stack byte
argument (a1 was a 5th param at sp+16). Gate MATCH whole-binary SHA-1
e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker C re-engaged on partition C closed 4 more (incl. 0x800450C4 which the
coordinator had recorded as a negative — C's source spelling beat the
coordinator's; the row is removed from the tracked index, the per-site
gp=-D_80121BFC override applied as recorded finding 16). 0x80065930 was
claimed by both C and the coordinator independently; skip-registered handled
the duplicate (coordinator's registration stood). Gate MATCH whole-binary
SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9. C's F15 (do/while
counter-init-first for bottom-tested search loops) and F16 (mirrored-layout
lever is per-GUARD, tested guard by guard) recorded for the cookbook.
Coordinator matched the 30-entry key/value pair-table lookup on the first
attempt (array spelling reproduced the two walked pointers). Gate MATCH
whole-binary SHA-1 e173426c157384ebf1b6caf8c6fea18a85a14af9.
Worker B measured that recorded negatives cluster at the head of every
partition: 39 of the first 40 worklist rows and 57 of 271 tier-1 rows (21%)
were addresses an earlier session had already attempted and recorded. The
triage tool never read the negatives index, so every closed negative
returned to the queue head and workers rediscovered them.
Fix: sf3_triage reads the tracked config/near_match_negatives.tsv (a new
--negatives input) and excludes those rows as recorded_negative=69. The
tracked index was grown from 38 Phase-8 rows to 94 by importing 56 Phase-9
negatives from the workers' reports and the coordinator's absorption record
(filtered to unregistered; tracked rows win on conflict). Worklist
regenerated: the head is now genuinely fresh (0x800F9134, 0x80042D88,
0x800450C4, ...). 3 new synthetic tests; plan-level tests wired for the new
argument. The phase plan said extract negatives only at close; B's
measurement showed the harm is per-cycle, so the index is now a living
tracked input. Note for P9-T8: keep importing worker negatives each cycle.
Christopher Williams