parallel_gate merged 17 across 16 binaries, aborted on a real red (ov_SC04_011). Reverted that one
file (func_80181F0C); the other 17 fns verify green from a full clean rebuild.
CORRECTION to my own first read: I attributed the ov_SC04_011 failure to func_80182DCC, whose agent
had flagged a needed jtbl carve for that same binary. The merged-and-failing draft was actually
func_80181F0C — I inferred from the binary name matching the carve note instead of reading the diff.
The carve-prediction pattern is still real and worth acting on, on the evidence that does hold:
three agents this session named, in advance, the integration reason the whole-binary gate would
refuse a draft they had just matched in isolation (func_801588CC's §8e/§260-A tail-class jtbl carve;
func_80182DCC's jtbl_801ED970 8-word carve in tail19.data.s; func_8001AF34's JTBL_PADS residual).
Those are gate-side work, not drafting failures, and must not be re-drafted — that is how
ov_SC06_022:func_80181664 burned three slots.
Hardest draw of the session: 23 of 25 targets over 120 instructions (up to 138), and the FIRST wave
whose packs auto-carried the residual class's cookbook bucket (15 of 25). Agents self-reported
25/25 MATCH with zero NEAR — against O21's 34/40 earlier the same day on an EASIER band. One wave
is not causation (the arm mix also moved toward opus), but the notes changed character: they cite
and EXTEND banked laws instead of grinding.
Gate refused 2, and BOTH were predicted by their own agents as integration rather than codegen:
func_801588CC's note named the §8e/§260-A tail-class jtbl carve (jtbl_80199FF8 parked adjacent to
.rodata) and said it needs the gate's jtbl lever, not C. func_8017E498 likewise. Reverted those two
binaries; the other 16 verify green from a full clean rebuild.
Headline lever from this wave (func_80181E98) is a SCOPE CORRECTION to §186, the same shape as the
§167-05 polarity fix earlier today: §186 says no C-level barrier steers cross-jumping — true for
STEERING, but a zero-byte __asm__ __volatile__("") before an arm's goto BLOCKS a suffix merge
outright. Traced to gcc's own jump.c: simplejump cross-jumping calls find_cross_jump(insn,
JUMP_LABEL(insn), minimum=1) and walks BACKWARDS from the j; the empty asm is an ASM_INPUT that
fails rtx_renumbered_equal_p at compare #0, so last1==0 and no merge occurs, for zero emitted bytes.
Closeness 75 -> 40 plateau -> MATCH.
parallel_gate merged 23 drafts across 22 binaries; its R22 caught 5 real failures
(ov_SC03_110:func_80180270, ov_SC06_025:func_8018098C, ov_SC03_024:func_8017DE8C+func_801838CC,
ov_SC02_037:func_801588CC, ov_SC07_010:func_8017EC6C). Reverted those five binaries' files; the
remaining 22 fns verify green from a full clean rebuild.
Lane conversion this round, same models and packs, drawn minutes apart:
O21 overlays 40 targets, 18-128 ins -> 34 self-reported MATCH (85%), 22 gated
M2 main 15 targets, 74-102 ins -> 4 self-reported MATCH (27%), 1 gated
Size does not explain it (M2's band sits inside O21's). Nearly every O21 MATCH cites an in-TU twin
or neighbour — overlays are ~134 near-copies of one engine, so a structural relative is usually
already banked. src/800.c is single-copy game code with no twin to lean on. Budget main at roughly
a third of the overlay rate, and reach for the permuter there rather than more drafting slots:
main's NEARs cluster at closeness 3-16, i.e. regalloc tail, not wrong code.
DeepSeek card-fueled drafts from .run/bakeoff/ds-fuel{A,B}. reloc_identity AGREE 10/10
pre-gate; whole-binary gate banked 9/10 (func_8017F468 @ ov_SC02_021 held back).
Parallel gate farm (tools/sweep_parallel.py -j 8) instead of the serial gate_lane:
7 binaries in 53s vs ~25min for the 19-binary serial sub-50 run.
The new head plumbing class after the symbol-kind fix: `conflicting types for func_80175414` (27
member-rows). Byte-true DEF is `void func_80175414(s32 _arg0)` (its DEFINE_ macro). The fleet
declared it 1,845 times in four spellings, of which three are the SAME TYPE (parameter names do not
participate) — the outlier was 28 sites declaring `(void)`.
conform_decls REFUSED the naive conform and was right to: 29 ZERO-ARG CALL SITES exist across 28
files, so conforming the declaration alone turns each into `too few arguments` — a fleet-wide
COMPILE break the per-binary gate cannot see (the tool cites 138/140 binaries, measured). It named
the count, the consequence, why the cheap check misses it, and the flag that repairs it, then
forced the two-step: --cast-zero-arg-calls (29 sites cast to the 0-arg fn-ptr shape, §17a-1 — gcc
folds the cast of a known symbol to a direct jal, so it is codegen-neutral), then the conform.
Result: 1,845 declaration sites rewritten across 1,061 files, 0 non-canonical remaining (axis
complete, R32). R22 clean-fleet: check-all 213 passed / 0 failed of 213.
WORTH RECORDING AS A TOOLCHAIN STANDARD: this is the instrument that has not wasted a cycle today.
Every other one reported SUCCESS over a defect — a classifier that discarded every gcc-2.7.2 hard
error (no `error:` prefix), a diff that miscounted 116 data-bundled .s files, a --verified-out
truncated to zero bytes over 62 real banks, a --band default that reported "0 families" on a real
135-member family, and a scope stamp describing the filesystem instead of the run. conform_decls
reports FAILURE with a repair path. A guard must state its COVERAGE, not just its verdict; the
in-repo exemplars are this tool and the §53 jr interlock.
Drew T