SS131: `sltiu N` is ground truth in BOTH directions. jtbl_range already EXTENDS a span the
dlabel cut short and WARNS when a span is shorter than the bound, but had no clamp for a
span too LONG for a NON-ZERO reason — and the trailing trim only removes ZERO words, so
ordinary data that spimdisasm ran into the dlabel slipped through and under-filled the piece.
Records the reusable FINGERPRINT of an under-fill, because it does not look like codegen:
hundreds of 1-byte diffs spread over most of the overlay, ~95% at byte 0 (mod 4) = the low
byte of a 16-bit immediate, every one changing by exactly -4. Bucket differing bytes by
offset%4 and decode a few words; uniform small deltas in the immediate field mean LAYOUT,
not codegen. (Measured: 812 of 853 at pos 0 mod 4, all -4.)
The clamp's authorization matches the extension path exactly: unambiguous sltiu bound only,
and REFUSE LOUDLY if any surplus word is a plausible code address.
This was the single instrument failure that survived SS125's retraction round — the one case
where "the tool is broken" was actually true. Now fixed, with the 710-ins behemoth banked.
The last surviving "compiler wall" of the session turned out to be an off-by-one-word
carve. Chain, all byte-grounded:
SYMPTOM jtbl_carve diverges on ov_SC06_018 after a BYTE-NEUTRAL jr_isolate_all.
SHAPE image -3 bytes; 853 differing bytes in 699 scattered runs; **812 at byte 0 of a
word** = the low byte of a 16-bit immediate, every sampled one changing by
exactly -4. Not a shift, not a pad: hundreds of `%lo` operands 4 bytes low.
CAUSE func_80191C50's own `sltiu $v0,$v1,0xC` names 12 entries; the emitted .rodata
holds 12 words; the carve reserved 13. The 13th word (0x3038200A) is ordinary
NON-ZERO data that spimdisasm ran into the dlabel span (the next dlabel sits
past it). Reserve 13 / supply 12 => the .rodata piece UNDER-FILLS by one word
=> every later symbol slides down 4 bytes.
WHY MISSED The trailing-trim's axiom is "0x00000000 cannot be a jump target", so it only
trims ZERO words. This surplus is non-zero, so nothing trimmed. The tool already
treats `sltiu` as ground truth - it EXTENDS a table spimdisasm cut in half, and
it WARNS when a span is too short - but had no clamp for a span too LONG.
FIX: an over-span clamp with the SAME authorization as the existing extension logic -
only when the `sltiu` bound is UNAMBIGUOUS (one bound; a multi-switch fn cannot say which
table owns which), and only when the surplus words are NOT plausible code addresses. If any
surplus word looks like a real entry it REFUSES LOUDLY rather than silently dropping one.
jtbl_carve: jtbl_801D3BA4: clamped 1 trailing NON-ZERO word(s) - func_80191C50's own
`sltiu 12` names 12 entries and the surplus is not code
Then: carve BYTE-IDENTICAL, draft spliced, **func_80191C50 (710 ins) BANKED**.
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
Retires the JTBL-CARVE-BREAKS-BYTES ledger class: the one instrument failure that survived
this morning's retraction round was REAL, and is now a named bug with a fix - not a wall.
Fleet 93.25% fn-count / 89.2% instr / 80.5% distinct; R22 140/140 (thirteen runs).
Nothing running, tree clean, lock FREE.
Records for the fresh session: the ordered resume list (T1 + T3 need /effort ultracode and a
WAIT for the toggle; #9 and T5 need Max), the JR-PAIR-IN-ONE-O0-OBJECT wall with its untested
SS81-step-1 escape, and the S28 ROI evidence that a 15-target wave bought 12 banks and +0.00pp
headline — so waves are only worth resuming against HIGH-REACH targets.
Flags the milestone reality plainly rather than leaving it for T5 to discover: 89.2% instr
against a >=95% bar, with the remaining volume in main + the 39 type-1 modules (P31 scope).
P30 realistically closes on the milestone's LEDGER branch, which is an explicit either/or in
the approved milestone — Drew's call, deliberately.
Adds the S28 HONESTY LEDGER: five wrong calls this session, each caught by an oracle, none
committed. The standing consequence is stated once, at the top of the handoff: only a full
clean-fleet R22 counts, and a tool's exit status is never the oracle.
Also removes a duplicated results section left by my own earlier checkpoint edit.
The last 133 members of the -O0 cluster's sweep residue, banked. R22 CLEAN-FLEET:
extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
METHOD NOTE, because it is the difference from the two failures before it: my hypothesis
(the 5-vs-133 split tracks which -O0 file the member lives in) was cleanly REFUTED —
ov_SC07_006/007/011 were sub-split by me TODAY into _o0c and banked anyway. Instead of
forming a fourth theory I staged ONE member and compiled it. Two blockers, each only
visible after the previous was cleared:
1. `conflicting types for S_8013BC7C` — the templated body carried a typedef TEXTUALLY
IDENTICAL to one in engine_types.h, and gcc-2.7.2 (C89) rejects even an identical
typedef redefinition. The exemplar TU includes only common.h; the new _o0c files pull
engine_types.h via engine_core.h. (harvest_verify already strips provided typedefs at
gate time via cdecl.strip_provided_typedefs, so this half was self-solving.)
2. `previous declaration of func_8013BC7C` — the SIBLING's own TU declares the templated
function divergently: SS57's self-decl class, lever = --normalize-self-decls.
One flag. 133 staged / 133 BANKED / 0 failed.
Running total for today's follow-ups: func_8013C08C 137/137 + this 133 = 270 members banked,
against one genuine integration wall (JR-PAIR-IN-ONE-O0-OBJECT). Every blocker in all three
was a DECLARATION-ENVIRONMENT problem, and none was visible without compiling and reading
the output — three wrong guesses where I theorised, zero where I read first.
I reported func_8013B83C + func_8013BD74 as "BOTH BANKED - BYTE-IDENTICAL". That was
WRONG. My in-loop gate ran `make extract && make build` without `make clean`, and it gave
a FALSE PASS. The clean rebuild does not link at all:
ov_SC01_077_o0.c:(.text+0x10f8): undefined reference to `$L105'
ov_SC01_077_jr_801588CC.o: undefined reference to `func_8013C938'
- the first is a local label from the C-emitted jump table;
- the second is a PREVIOUSLY-MATCHED cluster fn going undefined (the incremental build
reused objects that still satisfied it).
Reverted; R22 140/140; nothing lost, nothing was committed.
SS42b named the stale-object trap for a FALSE FAIL. This is its MIRROR: a FALSE PASS, on a
change that is not even linkable — false in the most convincing direction possible, a green
SHA. Anything touching config/ (carve, resegment, split) must be gated by a full
clean-fleet R22 before it is BELIEVED, let alone reported.
THREE WRONG CLASSIFICATIONS ON THIS PAIR, all corrected in the ledger:
CC1-FAIL-UNREAD -> it has no compile error at all
JTBL-PAD-SPEC-DRIFT -> I had carved ONE fn of a TWO-table span; carving both gives
pad spec [0,4,4] and the filter is satisfied
"both banked" -> the false pass above
Real class: JR-PAIR-IN-ONE-O0-OBJECT. Both bodies ARE byte-correct (match_one --o0 and
rtu_match --o0 both MATCH, 272 / 198 ins); the blocker is integrating TWO jr fns into ONE
-O0 object. Untested escape: SS81 step 1, isolate one into its own code subseg so each
object owns exactly one table.
Also records the diagnostic ladder that found it: diff the two BINARIES and bucket each
differing byte against the function's own vram range. 3,749 of 3,791 diffs were OUTSIDE the
function, first diff near the overlay START, image 57 bytes LONGER - the SS8 signature of
.rodata floating to the front. That fingerprint separates a codegen residual from a layout
effect in one build, and it is what finally redirected three wrong guesses.
SS129a: post-carve, rtu_match/match_one COUNT THE JUMP TABLE AS INSTRUCTIONS. For
func_8013BD74 it reported `mine=198, target=226, 206 mismatched` — and 226-198=28 is
exactly the table's entry count. A draft that verified cleanly BEFORE the carve reads as a
total mismatch AFTER it, and the number looks like deep codegen trouble. SS81 says a jr fn's
match_one MATCH is not a bank; SS129a says its post-carve DIFF is not a diff either. Let the
whole-binary gate arbitrate.
SS129b: NEVER commit a carve whose owner is still a stub. harvest_verify refuses a dirty
tree (SS97) and the carve dirties config/, so committing the carve to get a clean tree is
tempting — and it STRANDS the carve. jr_inventory refused instantly (R32: "carve ownership
is not 1:1 ... UNOWNED 0x801d828c"), blocking every later jr operation on that overlay.
Reverted; R22 140/140. The route for a jr fn is the INTEGRATED jtbl_family_bank (carve ->
extract -> remap -> gate per sibling in ONE uncommitted transaction). The SS81 hand-chain is
for diagnosis; as a banking path its two constraints contradict each other.
Both were mine, both caught by oracles before any lasting damage, both now documented.
Ledger: JTBL-PAD-SPEC-DRIFT (func_8013BD74, with the exact SS8e error) and CC1-FAIL-UNREAD
(func_8013B83C — the diagnostic is genuinely unread; say so rather than guess).
Fleet 93.21% fn-count / 89.1% instr / 80.4% distinct.
The SS81 chain's step 2, landed on its own so the bank can build on it: harvest_verify
refuses to run on a dirty tree (SS97 gate hygiene), and the carve necessarily dirties
config/. Carve applied -> byte-identical -> R22 140/140 -> committed. The draft bank
follows as its own step.
SS94 said treat a family 0/N as a TYPE-CARRY failure until proven otherwise. It was one.
MECHANISM (read from the source, not inferred - SS125 rule 1): E_13C08C was a MULTI-LINE
typedef at FILE scope. extract_unit's preceding-decl backscan walks back over
extern/comment/blank/typedef lines, but a multi-line typedef presents its CLOSING line
(`} E_13C08C;`) first, which matches none of those prefixes. The scan stopped there, so every
templated sibling received the body WITHOUT its type and all 137 failed to compile.
FIX = SS100 (prefer the DRAFT-LOCAL form): a type only one function uses belongs in its BODY,
where it is part of the unit by construction. Byte-neutral (a type declaration emits no code),
gated on ov_SC01_077 before and after.
family_sweep --hseq --only 0x8013c08c --band all -j8 -> 137 BANKED / 0 failed
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140
WHY THE DIAGNOSIS WAS REDONE FROM SCRATCH (R35): the first pass concluded "the type is not
defined anywhere" - because grep was SILENTLY SKIPPING the file (SS128, the raw-NUL defect fixed
at commit:1284). That conclusion pointed the same direction by luck, but it came from a broken
instrument and none of it was trustworthy. With grep working, the typedef was visible at
ov_SC01_077_o0.c:137 immediately.
A FRESH TRAP FOUND WHILE FIXING IT: my explanatory comment contained a literal `}`, and
extract_unit's forward brace-scan does NOT strip comments - it decremented depth and TRUNCATED
the extracted unit. Caught only because I verified the unit was complete instead of assuming
the edit worked. Comment reworded to contain no brace characters, with an in-place note saying
why. That hazard is now also documented in the body itself for the next reader.
FOUND BY ACCIDENT, WHICH IS THE POINT. `grep -rn func_8013C08C src/` returned NOTHING for
a function that is defined right there. The file held a RAW NUL byte inside a character
literal — the source read `== '<NUL>'` where it should read `== '\0'`. It COMPILES (the
fleet was byte-identical), so no byte-gate ever objected. But file(1) classifies such a
file as `data`, and **grep treats a file containing NUL as BINARY and reports nothing,
silently**. The whole file therefore vanished from every grep-based audit and every hand
search. I burned real time chasing a phantom missing function before `file` gave it away.
SCOPE, measured: 137 files — every `_o0c`/`_o0e` region created in THIS session. The
templated bodies carried the NUL fleet-wide, so I propagated the defect today. All fixed
(`'<NUL>'` -> `'\0'`); R22 CLEAN-FLEET 140 passed, 0 failed of 140 => byte-neutral.
NEW ORACLE: tools/audit_text_sources.py + `make audit-text-sources`, wired into
tools-health, coverage-asserting over all 3,887 tracked .c/.h files (R32). This is the
SAME silent-skip family as SS124 (a scanner that cannot see something reports it is not
there) and SS126a (a bare except swallowing a coverage assertion) — but one layer LOWER,
in the tool everyone reaches for first. The byte-gate is structurally blind to it (R34):
the bytes are correct, so it has nothing to say. It needs its own oracle.
MY OWN ERROR, RECORDED: proving the new guard fires, I injected a NUL into the REAL tracked
file and restored it through nested shell escaping. The restore left `'\\0'` — an escaped
backslash, i.e. a multi-character constant, NOT a NUL — a genuine semantic change. R22 caught
it (139/140) in one cycle, before any commit; repaired to `'\0'` (10465 -> 10466 bytes) and
re-verified 140/140. The lesson is not "be careful": a negative control must corrupt a SCRATCH
COPY under .run/, never the tracked file it is testing. Testing a guard must not risk
introducing the defect the guard exists to catch.
SS127/SS127a/SS127b distil what the wave's agents kept re-deriving, because the index
fired on only 3 of 15 targets:
- the -O0 CONSTANT-OFFSET FOLD: `p->f` folds to `lbu 3(r)`, `p[i]` does NOT (addiu +
0-displacement load). At -O2 these converge, which is why nothing in SS1-SS126 covers it.
- the -O0 regime generally: spill/reload pairs are REAL named locals; load-delay nops and
redundant copies are normal; write plain C, the -O2 steering levers are inert here.
- SS127a: SS71 sibling-first is the STRONGEST -O0 lever — an -O0 TU is a near-uniform code
regime, so a banked sibling's shape transfers far better than at -O2.
- SS127b: two agents' decisive levers came from a SOURCE COMMENT in ov_SC01_077_o0.c, not
from docs/. Promote levers out of source comments or every future agent re-buys them.
Checkpoint records the wave AND its honest ROI: 1.33M tokens for 12 banks and +0.00pp
headline. The value is contingent on three reach-138 functions, and all three are
currently unpropagated (func_8013C08C 0/137, SS94 type-carry) or gate-failed
(func_8013BD74 CARVE-REFUSED, func_8013B83C CC1-FAIL). Fix propagation before wave 2 —
drafting more x2-reach targets is not where the leverage is.
First Ultracode wave against the population the -O0 routing made draftable. 30 agents
(15 drafters + 15 adversarial verifiers), 1.33M tokens, 8.8 min. Every drafter self-checked
with match_one --o0 AND rtu_match --o0; every MATCH claim was then re-run from scratch by an
independent skeptic instructed to default to REFUTED. Result: 15/15 confirmed, 0 disputed.
WHOLE-BINARY GATE (the sole arbiter, G3/P9): **12 banked / 3 failed** — a textbook SS52b
outcome (an rtu MATCH is a CANDIDATE, not a bank). All three failures are NAMED INTEGRATION
classes, none a compiler wall:
func_8013BD74 CARVE-REFUSED — it is a jr function; needs the SS81 carve chain (reach 138)
func_8013B83C CC1-FAIL — real-TU compile, error not yet read (reach 138)
func_80184058 PLUMBING — recovery ladder
BANKED: func_8013C08C + the 11 fourth-region fns (func_80183CF0/D50/F28, func_80184028/264/
2E0/354/474/538, func_801847EC, func_80184868).
Bank truth read from the SOURCE (INCLUDE_ASM absence), never the gate report (SS55b trap 4).
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
PROPAGATION OF func_8013C08C (reach 138) IS NOT DONE: the first sweep returned "0 families"
because the family map still listed it as a stub — regenerated it (the documented
crack-wave-sweep-map-regen path), after which the sweep found 137 candidates and banked
**0/137**. Per SS94 a family 0/N is a TYPE-CARRY failure until proven otherwise, and this
body carries a SS100 body-scoped typedef, so that is the first hypothesis to test. Recorded as
open, NOT as a wall.
FLYWHEEL FEEDBACK (R16), the honest read: the cookbook index fired on only 3/15 targets.
Agents independently re-derived the SAME undocumented idiom — the -O0 CONSTANT-OFFSET FOLD
(`p->f` folds to `lbu 3(r)`; `p[i]` does NOT, it emits `addiu; lw 0(r)`) — and two found their
decisive levers in a SOURCE HEADER COMMENT in ov_SC01_077_o0.c rather than in the cookbook.
The -O0 regime is under-documented relative to how much of the frontier now lives in -O0 TUs.
SS71 also generalises to -O0: several agents cracked their target off an already-banked sibling
in the same TU (func_80184868 came straight off the shape banked earlier today).
(1) --fix-def-sig POSTURE: AUDITED CLEAN. `action="store_true"` (defaults False), one
consumer via getattr(a,"fix_def_sig",False), and NO caller anywhere passes it — checked
tools/, .run/ scripts, docs recipes and the Makefile. The flag help already carries the
SS119 warning.
BUT the audit surfaced a live hazard the earlier pass missed: docs/decision-log.md
still recommended "--fix-def-sig should likely be default-on for the h_seq path".
That was byte-REFUTED by T84/SS119 — the flag is a REPAIR, not a default; on 0x80161c98
it imposed a signedness-wrong `s32 a1` over the true `u32`, turned a byte-correct draft
into a 1-instruction DIFF (slti vs sltiu), and held 137 members at 0 until DROPPED.
Struck through in place with a superseding note rather than deleted, so the original
reasoning stays legible (R31) — but a forward-looking "should be default-on" sitting in
a doc a fresh session reads FOR DIRECTION is a hazard, not a historical note.
(2) GRINDER WARM-START: tools/permuter_ils.py has sat beside grinder.py since Phase 24 and
was never wired in, so every grind was a COLD search that burned its whole time box
re-descending ground the previous run had already covered. grinder.py now runs `--cycles`
(default 4) timeboxed permutes, each warm-restarted from the previous cycle's best byte
waypoint, stopping early on no gain. `--cycles 1` reproduces the old cold behaviour exactly,
so it is opt-out. --permute-secs is now documented as the PER-CYCLE box.
JUSTIFIED BY MEASUREMENT, not by the task list: the lane looked dead (Phase-22 audit: 7
all-time banks, all Phase 21, 0 since), so I checked for live fuel before building. The
backlog holds 665 open near-misses in the permuter-tractable band (close 1-20), 157 of them
close 1-4, including func_8016BA68 at close=1 with reach=134.
HONEST LIMIT: this is a WIRING change whose yield is UNPROVEN. The Phase-24 evidence for ILS
is one function (func_80148094, 72 -> 36 over ~8 restarts); I have not run it on this
backlog. A winner remains a CANDIDATE — the whole-binary byte-gate is still the sole arbiter
(G3/P9), and an intermediate waypoint is only ever re-seeded, never banked.
The second and last has_mid_jr family of the -O0 cluster. Same route as func_8013C0F8:
jtbl_family_bank (the SS81 carve chain per sibling) -> 137 BANKED / 0 failed.
Both jr families together: 274 members, 483 ins x137 = ~66k instructions, 0 failures.
Neither was bankable before commit:1270 routed the destination TUs to -O0.
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
The first of the two has_mid_jr families SS53 correctly refused from the carve-less
family_sweep. Routed through the tool its tier needs (jtbl_family_bank, the SS81 carve
chain per sibling) it banks CLEAN:
jtbl_family_bank func_8013C0F8 ov_SC01_077 0x8013c0f8 -> 137 BANKED / 0 failed
~7s per sibling (carve + extract + remap + whole-binary gate), ~16 min total
Only possible now because commit:1270 routed each destination TU to -O0; before that the
member's home file compiled -O2 and no body could ever match there (SS116).
INDEPENDENT CONFIRMATION OF THE SS125 DIAGNOSIS: this is the SAME tool that returned
gate-fail on every group-B (func_8017BEBC) probe earlier today. 137/137 here versus 0/4
there, same jr machinery, is exactly what the carve-vs-body diagnostic concluded --
group B's failures are its BODY, not the carve. I had first blamed the carve; the
body-free probe corrected it, and this run corroborates the correction from the other side.
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
The payoff of routing the cluster to -O0 (commit:1270). These functions were ALREADY
CRACKED in ov_SC01_077 and could not be banked anywhere else purely because every
destination file compiled -O2. With the destinations now -O0, they template in
deterministically -- no drafting, no agents.
dedup_propagate --recover 0x8013C360 (h_exact x138) -> 137 overlays byte-identical
family_sweep --hseq 10 variant families -> 1,227 banked / 133 failed (90%)
1360 staged across 136 groups
------------------------------------------------------------------------------------
1,364 new banks
FLEET: fn-count 92.71 -> 93.09% · instr 88.3 -> 88.6% · distinct-code 78.7 -> 79.3%
(71,756 / 87,459 unique fns; +1,162 unique). dedup 1904 -> 1905 groups, 0 failed;
C1 coverage 240496/240496. 0 NON_MATCHING in any default build (G4).
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
--recover WAS LOAD-BEARING (SS75): without it dedup_propagate took its historical
all-or-nothing branch -- one failing overlay (the SOURCE, ov_SC01_077) dropped the whole
function and it printed "all candidates dropped", which reads exactly like a wall. Reading
the exclusion code instead of believing the message showed the remedy: --recover excludes
only that overlay (kept x1 with its own inline match) and propagates to the other 137.
The two has_mid_jr families in the cluster were REFUSED BY DESIGN, not attempted (SS53
interlock): 0x8013C0F8 (154 ins) and 0x8013C414 (329 ins), ~137 members each = ~466
members queued behind the jtbl carve path they actually need, rather than a fake 0% from
the wrong tool.
REMAINING in the cluster: the 133 sweep failures + the 2 jr families + the 3 addresses
never cracked anywhere (0x8013B83C, 0x8013BD74, 0x8013C08C) -- the last are genuine
drafting work, now finally possible since their TU is -O0.
I under-counted this cluster 8x (reported 275 stubs/18 overlays; truth 2,184/138). The
scan ran during a background rebuild AND wrapped corpus.stubs() in `except: continue`,
so every R32 coverage refusal became a silent skip and the total was taken over the few
overlays that happened to be re-extracted already.
Two of our own rules broken at once: a measurement taken during a rebuild is not a
measurement (caught EARLIER the same session, by the same assertion I then suppressed),
and R32 lives in the CALLER — an oracle only asserts coverage if the caller lets it raise.
It also cost credibility the other way: I used the bad number to declare the T0(f)
"2,192 open members" pin STALE. The pin was right. R35 applies to a re-measurement as
much as to the original measurement.
Checkpoint updated with the corrected population and the completed fleet-wide sweep.
Applies tools/o0_subsplit.py across every overlay whose 0x8013B568..0x8013C98C -O0
cluster was trapped inside an -O2 jr split. This is the population the phase opened
against, and it has never been buildable-at--O0 before.
135/135 sub-splits applied, 0 tool refusals
140 new -O0 region files; corpus.o0_sources() 137 -> 277
0 of them invisible to the -O0 oracle (verified explicitly -- a SILENT -O0 miss is
the exact failure SS126 warns about: the region would compile -O2 and every residual
it produced would be a pure artifact)
**2,200 open stubs now live in a genuinely -O0 translation unit**
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
make tools-health OK: corpus(+resident) 0 PHANTOM/0 TRUNCATED; cdecl ALL ORACLES GREEN;
audit-binaries 140 onboarded, every one a full citizen (R36); dedup-check 1904/0,
C1 coverage 240359/240359; cookbook-index 336 sections.
The transform is byte-neutral by construction (it only moves subseg boundaries and
repartitions source), so the whole batch was gated by one clean-fleet R22 rather than
135 individual builds -- after a single-overlay probe (ov_SC01_000) proved it (R37).
NOTE THE POPULATION CORRECTION (R14, mine): I earlier reported this cluster as "275 open
stubs across 18 overlays". That was WRONG -- the sizing scan ran while R22 was rebuilding
in the background, so corpus.stubs() raised for most overlays and my bare `except:
continue` SWALLOWED the very coverage assertion R32 exists to raise. The true figure is
2,184 open stubs across 138 overlays, which vindicates the T0(f) pin of "2,192 open
members" that I had called stale. The target list for this sweep was rebuilt with NO bare
except, so R32 can do its job.
Drafting fuel confirmed present: cached Ghidra-C seeds exist for the cluster's addresses.
Drafting the 2,200 is crack-wave work (T3), not T2.
- T0.5 was COMPLETE in SESSION-27 (124/124 programs, 7,716 files) but left unticked;
corrected, with the omission noted rather than silently fixed.
- T2 marked [~] part-done and its DEVIATION recorded against the phase-start plan:
the PRIMARY (two-file atomic o0b append) was byte-refuted, and the FALLBACK premise
was ALSO wrong — Arm-A does not bite. What actually blocked it is §126. Route proven,
tool shipped, 6 banked; the 18-overlay sweep of the 0x8013B568 cluster remains, sized
at 275 open stubs (re-derived; the T0(f) "2,192" pin is stale).
Records the T2 result as the phase's biggest unblock: the carve-within-a-carve is
byte-neutral (Arm-A does NOT bite), the real constraint is that an address range is
not an optimization region (SS126), and tools/o0_subsplit.py implements the correct
bound. Measured, not assumed, what it unblocks: 275 open stub instances across 18
overlays in the 0x8013B568..0x8013C98C cluster, homed in an -O2 jr split — plus a
note that the T0(f) "2,192 open members" pin is STALE and must be re-derived before
costing (R37).
Also flags my own under-count: the "15 contiguous -O0 fns" came from an asm scan that
cannot see matched functions.
Promotes the proven probe (commit:1266) into a real tool, and validates it FIRST-TRY on a
fresh overlay.
tools/o0_subsplit.py <ov> --lo <vram> --hi <vram>:
- derives the range's contents from the SOURCE ANCHORS (overlay_src_split.parse_overlay_c:
`asm` = unmatched stub, `define`/`def`/`nonmatch` = already matched), NEVER from an asm
scan -- a matched fn emits no .s, which is exactly the blindness that made the range look
like a clean contiguous run (SS126 / SS124's shape);
- computes the -O0 bound as (address range MINUS already-matched bodies) and emits ONE
sub-region per maximal run of unmatched anchors (K matched islands => K+1 regions);
- names each `<ov>_o0<letter>` picking free suffixes, so the widened Makefile glob selects
them; refuses loudly if it runs out or if the range spans >1 object or is already -O0;
- honours the one-carve-per-region law (forces a cut at every already-banked jr in the
object) and reuses jr_isolate_all's plan/build_new_config/ascending-unique validation
verbatim, so carve-repoint + source-repartition stay on the proven path;
- warns (does not refuse) when a stub in an -O0 run lacks the frame-pointer prologue --
the byte-gate is the arbiter, not the heuristic.
VALIDATION on ov_SC03_015 (untouched by the manual probe): the tool independently derived the
SAME structure found by hand on ov_SC03_014 -- 2 matched -O2 islands (func_80184440,
func_801848E4), 2 -O0 regions (8 + 7 fns), same 5 cuts. Sub-split -> BYTE-IDENTICAL. Then 3
drafts, each global DERIVED FROM THAT OVERLAY'S OWN ASM (%hi operand) rather than copied:
3/3 match_one --o0 MATCH (22 ins), 3/3 through the whole-binary gate.
BANKED this commit: func_801846E4 / func_8018473C / func_80184794 in ov_SC03_015 (6 across
the two overlays now). The other 24 stubs in the region are undrafted -- the route makes them
DRAFTABLE (they were un-bankable at any effort before); drafting them is crack-wave work.
R22 CLEAN-FLEET: extract-all 139/139 (+main); check-all 140 passed, 0 failed of 140.
cookbook SS126 (the address-range-is-not-an-optimization-region law + the probe ladder).
T2's central unknown is resolved, and it is NOT what the phase plan predicted. The plan
named the Arm-A splat `%lo +0x20` defect as T2's real substance. Four probes, each
isolating ONE variable, SHA vs config/check from a clean tree (SS125 rules):
1. sub-split the jr object at ARBITRARY addresses, pure -O2 -> BYTE-IDENTICAL
** the re-carve is NEUTRAL; Arm-A does not bite here **
2. same split, middle region routed to -O0 -> DIVERGED (2 vars at once)
3. same NAME as probe 1, only the -O0 flag added -> DIVERGED
** therefore the FLAG, not the subseg name **
4. -O0 regions cut to EXCLUDE the matched bodies -> BYTE-IDENTICAL
** route PROVEN end-to-end **
THE REAL OBSTACLE: an address range is not an optimization region. Interleaved among the
15 -O0 stubs at 0x80183CF0..0x80184920 are TWO already-MATCHED functions (func_80184440,
func_801848E4) that expand from engine_core.h and compile at -O2. Flipping the FILE to -O0
recompiles them too. Cut around them and it is byte-clean.
MY EARLIER "15 contiguous -O0 fns, clean cut" WAS AN UNDER-COUNT (R14 on myself): I derived
it by scanning asm/**/*.s for the frame-pointer prologue, and a MATCHED function emits no
.s -- so the scan was structurally blind to exactly the bodies that break the flip. Same
shape as SS124. Any T2 driver must derive -O0 bounds as (address range MINUS already-matched
bodies), never from an asm scan.
BANKED (3, whole-binary gate the sole arbiter): func_801846E4 + siblings func_8018473C /
func_80184794. Each global DERIVED FROM THE ASM (%hi/%lo operands), not taken from the
draft's comment; 3/3 match_one --o0 MATCH (22 ins) with a -O2 control showing the mismatch;
3/3 verified through harvest_verify; bank truth read from the SOURCE (SS55b trap 4).
MAKEFILE: the -O0 glob widened `ov_*_o0b.c` -> `ov_*_o0?.c` so ANY lettered -O0 sub-split is
covered by one rule. A MISSED rule is silent -- the region would compile -O2 and every
residual it produced would be a pure artifact (SS116). corpus.o0_sources() re-verified: 137
sources, resolves `?` via glob, both pre-existing rules intact.
CONFIG: ov_SC03_014_jr_8017EB7C sub-split into 5 regions (pre / _o0c / matched-O2 /
_o0d / post), reusing jr_isolate_all's plan+build_new_config+validation verbatim so the
carve-repoint and source-repartition semantics are the proven ones.
R22 CLEAN-FLEET: extract-all 139/139 (+main) ; check-all 140 passed, 0 failed of 140.
Max-effort re-measurement of the three jr refusals I ledgered earlier this session.
Two of the three verdicts were FALSE. Every number below is SHA vs config/check.<ov>.sha
from a clean tree, with the restore re-verified.
func_8018057C / ov_SC01_009 : jr_isolate_all is BYTE-NEUTRAL
-> "JR-ISOLATE-BREAKS-BYTES" RETRACTED; original failure not reproducible.
func_80191C50 / ov_SC06_018 : isolate NEUTRAL -> carve DIVERGED
(got 1b1667ea, want cbbc4f44) -> the ONE real instrument failure. CONFIRMED.
func_8017BEBC / ov_SC04_004 : carve is BYTE-NEUTRAL (body-free)
-> failure is the TEMPLATED BODY, the OPPOSITE of what SS125 first claimed.
Re-probed once more from a verified-clean tree: still gate-fail. Reclassified
BODY-TEMPLATE-GATE-FAIL.
So the tidy "two apparent walls are ONE tooling problem" conclusion was wrong: they
are two different problems, and the third target has no demonstrated problem at all.
ROOT CAUSE, and it is mine not the tools': a grep-of-the-build-log gate inside a driver
that did not revert on abort. config/overlays.mk is SHARED, so target 1's half-applied
isolate was still in the tree while target 3 was measured. Separately reproduced the
SS42b stale-object trap head-on: `git checkout -- config/` WITHOUT a re-extract turned a
byte-identical overlay into [FAIL] got 8f28aa77 / want 38a3d919 (Phase-20's R22
corollary, live).
SS125 rewritten. The METHOD (split the carve from the body, one build) is kept and is
what refuted this section's own first conclusion; what is added is the instrument rules
that make its answer trustworthy: compare the SHA against config/check, never grep the
log; re-extract after every config change AND every revert; a driver that aborts a
target must revert it before the next; verify the BASELINE against canonical too.
Meta-lesson recorded: SS53 says a 0% from the wrong TOOL manufactures a doctrine — this
is the same failure one level up, a verdict from the wrong MEASUREMENT, and my own
diagnostic script is an instrument subject to R35 like any other.
Ledger corrected in place (3 entries, superseding the earlier misattributions), so the
scheduled repair is the right one. No source/config change; no bank affected; the fleet
is untouched at 140/140 (last full R22 this session, HEAD commit:1263).
The session's most useful finding is an instrument ticket, not a match.
SS125 (new): before ledgering any jr residue, run jtbl_carve with NO body spliced
and rebuild. Byte-identical => the carve is neutral and the failure is the template;
NOT identical => the failure is the carve and the body was never fairly tested.
One build, and it collapses ambiguity that SS53 warns has twice steered strategy.
MEASURED: group B func_8017BEBC had gate-failed 3 probes in a row (default AND
--raw, cross-address ov_SC02_015 AND same-address ov_SC04_004). Carve-only on
ov_SC04_004 broke the bytes with nothing spliced — so all 3 probes were testing a
body that never got a fair run. The SAME stage had already refused behemoth
func_80191C50/ov_SC06_018. Two "unrelated walls" = ONE tooling problem.
func_8018057C/ov_SC01_009 fails at a DIFFERENT stage (jr_isolate_all, step 1) and
is deliberately NOT grouped with them.
Both tools reported SUCCESS on every failing target; only the whole-binary gate
refused (G3/P9). A tool's exit code is not the oracle.
Ledger: the three logged by STAGE (JTBL-CARVE-BREAKS-BYTES / JR-ISOLATE-BREAKS-
BYTES), not by function, so a carve fix auto-reopens every target it should.
None is diagnosed, so none is called a compiler wall — that guess has been wrong
four times running on this project (R35).
CURRENT_PHASE: SESSION-28 checkpoint refreshed; the carve diagnosis is now resume
item 1 (it gates 13 members + a 710-ins behemoth and every future jr family).
The 3 jr behemoths carried from SESSION-27 (staged .run/beh-gate/), each run through
the SS81 3-step chain SERIALLY under tools/treelock.sh (every step calls `make extract`).
Re-derived the stub set on a QUIESCENT tree first: 4 of the 7 staged fns were already
banked in S27 and the staging dir still held all 7.
BANKED (1):
ov_SC05_010 func_80181CDC (769 ins) — jr_isolate_all --only -> BYTE-IDENTICAL,
jtbl_carve --func -> BYTE-IDENTICAL, then harvest_verify --chunk 1. Bank truth read
from the SOURCE (INCLUDE_ASM absence), never the gate report (SS55b trap 4).
BYTE-REFUSED (2) — to the wall ledger WITH EVIDENCE, not forced, not labelled walls:
ov_SC01_009 func_8018057C — jr_isolate_all reported success (2 jr in 1 -O2 object,
2 region .c) but the post-isolate build was NOT byte-identical. Aborted at step 1.
ov_SC06_018 func_80191C50 — isolate clean; jtbl_carve reported success (48-piece
carve set, 22 tail pieces) but the post-carve build was NOT byte-identical.
Aborted at step 2.
Both TOOLS claimed success; the whole-binary gate refused (G3/P9 doing its job).
Neither is diagnosed, so neither is called a compiler wall — on this project that
guess has been wrong four times running (R35). Drafts + logs preserved under .run/.
R22 CLEAN-FLEET (config changed => T2 blast radius => mandatory):
extract-all: 139 extracted, 0 failed of 139 (+ main, serial)
check-all: 140 passed, 0 failed of 140 <- BYTE-IDENTICAL
DRIVER DEFECT FIXED (mine, SS105/SS61): v1 of .run/s28_beh.sh `continue`d past a failed
gate WITHOUT reverting, leaving two half-carved overlays in the tree; only a full
`git checkout -- src/ config/` recovered it. Nothing was ever committed, so the cost was
build cycles and zero work. v2 reverts that target's config+src on EVERY failure path.
- §124: a "not matched" verdict can mean the definition is there under a DIFFERENT
C NAME (the §37/§73 asm-label alias). The whole "no matched unit" skip class was
one exemplar x 137 members. Includes the two traps in the fix (re-derive the
pattern PER FILE; CARRY the alias declaration or the sibling emits the wrong
symbol and still links) and the law: when corpus.stubs and a source scanner
disagree, the SCANNER is wrong.
- §124a: `0 matched-exemplar families` from family_sweep may be the --band FILTER
(defaults to `substantial`), not a wall — same shape as §53 / §116.
- cookbook-index regenerated (tools/cookbook_index.py, R33).
- CURRENT_PHASE: SESSION-28 checkpoint. Fleet 92.70 fn-count / 88.3 instr / 78.7
distinct, R22 140/140. Records the 4th -O0 region VERIFIED + SIZED (30 instances
/ 1,504 ins, ov_SC03_014+015 only) and correctly BLOCKED on the T2 re-carve, and
two R14 corrections to my own SESSION-27 checkpoint (the 137 were skips not
failures; the cause was not "banked in the wrong binary").
The skip class that the SESSION-27 checkpoint carried as "~137 members, a cheap
sweep-routing gap" is closed: ONE exemplar, ALL 137 same-address members banked,
0 failed, in a single --only sweep after the extract_unit alias fix (commit:1260).
- family_sweep --hseq --only 0x8016191C --band all -j 8, under tools/treelock.sh
(the campaign-wide mutex, not a pgrep poll).
- BANKED 137 member-matches / 0 failed across 137 overlays; skipped {}.
- 24 ins x 137 = 3,288 instructions that were sitting behind a tool lookup miss.
- R37 probe before the sweep: rtu_match MATCH (24 ins) on 3 members in the REAL
TU (ov_SC01_000 / _001 / _004) before spending 137 gate cycles.
R22 CLEAN-FLEET: make clean && make extract-all && make check-all ->
extract-all: 139 extracted, 0 failed of 139 (+ main, serial)
check-all: 140 passed, 0 failed of 140 <- BYTE-IDENTICAL
0 NON_MATCHING in any default build (G4). The whole-binary byte-gate was the sole
arbiter for every one of the 137 (G3/P9).
Note --band: the sweep defaults to band=substantial and this family is band=mid,
so the first invocation returned "0 matched-exemplar families" — a silent-looking
zero that is a FILTER, not a wall. --band all is required for mid/tiny families.
The h_seq sweep's 137 "no matched unit for func" skips were ONE exemplar x 137
same-address members, not 137 distinct failures: func_8016191C @ ov_SC01_077,
band=mid, non-jr, 24 ins, ALL 137 members still INCLUDE_ASM stubs => 3,288 ins
left on the floor by a tool lookup miss (R35 again).
ROOT CAUSE (and it refutes the SESSION-27 checkpoint's own diagnosis, R14): the
exemplar is NOT "banked in a different binary". It is banked in ov_SC01_077 under
the SS37/SS73 asm-label alias — the byte-true body conflicts with the fleet-canonical
decl on BOTH SS73 axes (return void vs s32 AND params void*/s32 vs int/unsigned),
so it was banked zero-touch as
int aF8016191C(int, unsigned int) __asm__("func_8016191C");
extract_unit only ever matched a definition head literally NAMED func_<ADDR>, so
it returned None, and every caller reads None as "not matched".
FIX (tool-only, T0 blast radius):
- _alias_decl_for(): resolve `<ident>(...) __asm__("func_<ADDR>");` -> <ident>.
- extract_unit(): accept the alias identifier as the definition head, re-derived
PER FILE so one file's alias can never leak into the next.
- carry the alias DECLARATION into the unit (without it the sibling emits the
symbol aF8016191C and the body never lands at func_<ADDR>); the backscan now
walks PAST the alias line so the fn's own preceding externs are carried exactly
as for a plain definition, with a start<=alias_ln<=end guard against double-emit.
- R32: an alias decl with no findable definition now REFUSES LOUDLY instead of
falling through to _macro_unit and reporting "not matched" — the silent-skip
class this fix exists to delete.
Rejected the alternative the banking agent suggested (widen engine_core.h's decl
void->s32, drop the alias): it addresses only SS73's RETURN axis while decl and body
also disagree on PARAMS, and it is a T2 fleet-shared edit where the alias is T0.
VERIFIED: unit extracts with externs + exactly one asm label; remap_hseq produces a
sibling draft; rtu_match 3/3 MATCH (24 ins) in the real TU (ov_SC01_000/001/004).
make tools-health RC=0. The whole-binary byte-gate remains the sole arbiter (G3/P9).
The 12 agents killed by the usage-limit pause were resumed and ALL returned MATCH (2 had already
banked from their partial drafts, so 10 ran). h_exact propagation leg completed over all 112 banked
exemplars: 14 propagated, 42 benign skips (h_seq tier, correctly routed away per §123), 0 failures
— the 0x801466F0 'halt' was a third benign-refusal phrase, not a partial write.
fn-count 92.61 -> 92.67% | instr 88.2 -> 88.3% | distinct 70,581 -> 70,590 unique fns.
Every line is a symptom an agent HIT and had to re-derive from gcc internals because title-keyword
search structurally cannot surface it:
- SIZE-MISMATCH/short + frame-pointer prologue => the target is -O0, pass --o0 (the flag was
documented nowhere an agent would look; a 4th -O0 region also exists beyond the 3 known ones)
- rotated instruction window => sched1 order; brute-force all N! statement orders (24 runs, 2 min)
- if/else result in $v1 vs target's $v0, and load-hoisted-above-store => §76 variable reuse
(§76's title reads behemoth-only, so nobody finds it for a 48-ins function)
- ori 0xffd8 vs addiu -0x28 => negative const in an UNSIGNED narrow local; signed keeps the lhu
- LENGTH-DRIFT -1 as a missing jal-delay copy => narrow ANSI prototyped param (not just K&R §43)
- lwl/lwr+swl/swr is a delay-slot SPONGE (the inverse of the §5a fence case)
- a vanished param copy => cse.c make_regs_eqv live-range rule
- a ghidra_c seed may be a DIFFERENT function (overlays share VAs)
WAVE 3 (48 agents / 8 binaries, dealt across binaries so BANKING fans out): 48/48 match_one MATCH,
48/48 banked through 8 PARALLEL per-binary gates. WAVE 2: 15/19. BEHEMOTHS: 4 non-jr confirmed
(func_8017E120 884ins x14, func_8017FA5C 728, func_8017CAD4 755, func_8017E35C 719).
Tier-routed propagation (§123): family_sweep --hseq banked 911 members across 137 overlays.
fn-count 92.32 -> 92.59% | instr 87.9 -> 88.2% | distinct 78.3 -> 78.7% (70,506 unique fns)
R22 clean-fleet 140 passed / 0 failed, under one campaign lock (treelock.sh).
CORRECTION (R14): the 'per-binary bank-rate cliff' I reported from the pre-incident gate run
(SC03_014 1/6, SC04_018 1/6, SC06_018 2/6) was an ARTIFACT — those gates ran against a tree
propagation was concurrently rewriting. Re-gated clean: 6/6 everywhere. A measurement taken during
corruption is not a measurement; I should not have theorised a cause before re-running it.
I gated 8 binaries in parallel while wave-2's propagation loop was still running, then ran
'make clean' on top. check-all 77/140; the corpus denominator moved, so the apparent 91.4% instr
was a half-written tree, not a gain. Reverted to commit:1245 (last R22-verified) — 140/140 restored,
all 58 drafts survived because agents only ever write .run/.
ROOT CAUSE, and it was structural not unlucky: my guard was
while pgrep -f dedup_propagate; do sleep; done
A CAMPAIGN is a LOOP of short-lived processes (15 sequential invocations), so it has gaps where no
process matches. The poll sampled a gap and started. Presence-of-a-process cannot express 'a
campaign owns the tree'.
treelock.sh holds one flock for the WHOLE campaign, released by the kernel on exit OR kill, with
--status; both drivers refuse to run unlocked. LAW: guard the CAMPAIGN, not the process.
Corollary (twice today): a killed process performs no undo — a fleet-tier write needs a lock ABOVE
it, not cleanup inside it.
run_gate() always took per-worker result paths; the CLI never exposed them, so every CLI gate used
the shared .run/harvest_{verified,failed}.txt. Concurrent per-binary gates — safe on every other
axis, since the byte-gate IS per-binary — would have read each other's results and mis-attributed
banks (the §55b trap-4 shared-scratch defect that bit match_one in P28, one level up). Default is
now .run/harvest_verified.<binary>.txt: safe by construction, not by remembering a flag.
WAVE-2 MEASURED THE INDEX: 15/19 index hits and the bank rate went 57% (wave 1, no index) -> 79%
(wave 2, index-first) on the same gate. Agents also NAMED its gaps, which is the flywheel working.
Two real defects found and fixed:
1. COVERAGE. The parser required a '§' prefix, so 111 h2-h4 headers were invisible — including
'### T4 — Branch polarity', the fix match_one names by class (BRANCH-POLARITY) and which two
agents re-derived by hand, and the §1/§2 idiom-catalog entries (I1-I4, T1-T4).
2. THE ASSERTION ITSELF. My R32 check compared §-headers-parsed against §-header-CANDIDATES — a
tautology over a set I had already narrowed. R32 says the candidate set must OVER-approximate;
it now counts EVERY header and accounts for each as indexed-or-explicitly-skipped. The tool
written to stop silent skips had the silent-skip defect.
3. Keyword matching over titles cannot surface an idiom whose title omits the symptom, so the
index now opens with a hand-curated SYMPTOM -> section list, seeded from what agents actually
hit (branch polarity, (void)-canon conflicting types, asm-label alias, one-base-register reuse,
folded andi, slti/sltiu, sibling-first, delay-slot theft, void->s32).
Verified against code rather than the ledger: the split-blind lookup globs */ correctly, and the
churn was fixed by T5's input-signature gating. Both struck. asm_subdir_for was still a parallel
oracle (silent g[0] on multi-match) -> now corpus.asm_path. --fix-def-sig defaults off correctly
but advertised 'Byte-neutral; gate arbitrates' — the claim T84 refuted (signedness-wrong header
decl over a byte-correct draft; 137 members held at 0 until the flag was dropped).
A defect ledger nobody re-verifies decays into busywork — verify before scheduling.
Wave-1 measured the tax: three agents each reported a 'NEW idiom' that was ALREADY documented —
the asm-label alias (line ~2516, same 'address-of perturbs regalloc' mechanism) and the void->s32
non-neutrality (§41d, Phase 26; the agents cited the very entry §41d corrects). They consulted the
cookbook as instructed and could not FIND them. 716 KB / 226 sections with no index = a
discoverability failure, and every wave re-paying for prior waves' findings is the inverse of R16.
docs/cookbook-index.md maps SYMPTOM (what you see in the diff) -> sections, 14 buckets, a section
listed under every symptom it addresses. Derived by tools/cookbook_index.py (R33 — cannot drift),
--check wired into tools-health.
R32 on my own tool: the first regex required an em-dash separator and silently dropped 50 sections
— including §1 (idiom catalog), §2, §5a (cross-jump, cited by an agent today). An index missing its
most-cited entries turns 'I could not find it' into 'it is not there'. Now asserts extracted ==
candidate '§' headers and hard-exits on a gap.
The 4 cores dedup_propagate refused (h_exact tier) templated cleanly via family_sweep --hseq once
the family map was regenerated post-bank (a bank invalidates the map: sig-overlays + family_hseq
must run BEFORE the sweep — the standing wave-loop order). 548/686 banked, 138 failed (one
consistent per-overlay slice, diagnose next). Wave-1 total: 8 cores -> 1,121 instances.
instr 87.5 -> 87.9%, distinct 69,828 -> 70,094 unique fns.
Drew T