# Cookbook index — find the entry by the SYMPTOM you are looking at > **Generated by `tools/cookbook_index.py` — do not hand-edit** (R33). Regenerate after adding a cookbook section. > > `docs/matching-cookbook.md` is ~716 KB / 1176 sections. Grepping it blind is how three P30 wave-1 agents each "discovered" an idiom that was already written down. **Start here, then read the section.** A section appears under every symptom it addresses. **How to use:** name what you SEE in the diff (a stolen delay slot, an extra `la`, a swapped register pair, a `conflicting types` error), find that symptom below, read those sections first. If nothing fits, THEN grind — and add a section when you win. ## Start here — the symptoms that come up most, with the section that fixes them > Hand-curated from what agents actually hit and, in several cases, RE-DERIVED because keyword matching alone did not surface the entry (P30 wave-2 feedback). If your symptom is here, read the named section before anything else. - **wrong branch sense / arms swapped / `beq` where the target has `bne` (match_one prints this class as **BRANCH-POLARITY**)** → **§3-T4** (invert the source condition) + **§32**.2; for a trailing `return 0` vs an early return see the shared-ret0 note (cookbook L1344) - **`conflicting types` on YOUR OWN function's definition, where the fleet canon is `(void)`** → **§73** (the PARAMS axis) + **§42** — keep the `(void)` signature and read the incoming arg via `register s32 a0v __asm__("$4")`; §42 is otherwise indexed only under regalloc - **a shared global declared at a conflicting type (u8 vs s32, signedness) blocking your draft** → **§37** asm-label alias `extern T X __asm__("D_x")` — beats `*(T*)&X`, whose address-of perturbs regalloc. Works for FUNCTION definitions too (P30 wave 2) - **target reuses ONE address register across two different offsets of the same global** → **§20** (the pointer-var-to-the-global bullet, cookbook L1907-1918) — take `T *p = &D_x;` and index off `p`, never the bare symbol twice - **an extra `la` / the address hoisted into a callee-saved register across calls** → **§20** + `gcc-2.7.2-map/cse_expr.md` §H — `*(T*)&sym` force_regs the address; the asm-label alias (§37) keeps the direct `%lo` mem form - **`andi $x,0xFF` folded away in your output but present in the target** → **§1/I2** + **§12** name the family; if the prescribed `& 0xff` at the use folds, hold the masked byte in a **u16** local so only a QI->HI extend survives (P30 wave 2, byte-tested) - **`slti` where the target has `sltiu` (or vice versa)** → **§35** — a separate SIGNED int copy of an unsigned load keeps `slti`; chained bounds get range-folded, so write each bound as its own `if`/`goto` (**§21**) - **you are about to hand-derive a body that some overlay already matched** → **§71** — grep the callees/globals for an already-matched SIBLING first; in wave 2 this alone produced iteration-1 MATCHes on 4 of 19 targets - **gcc stole an instruction into a branch delay slot that the target leaves as `nop`** → **§5a** + the zero-byte `__asm__("")` fence (**§34** toolkit) — `reorg.c stop_search_p` halts the eager filler on an asm insn - **`void` vs `s32` return — is promoting it byte-neutral?** → **NO, not always: §41d** (byte-proven; a `void` body with no `return` gains an instruction). P30 adds a second mechanism: an `s32` return keeps `$v0` live-out and blocks dbr from filling a loop-back delay slot - **`match_one` says **SIZE-MISMATCH/short** and the target has a frame-pointer prologue (`addu $fp,$sp,$zero` / `21F0A003`)** → **The target is -O0.** Pass `--o0` to `match_one`/`rtu_match` — nothing else will ever match (§116: opt level is a property of the FILE). Known -O0 regions: boot, `ov_SC01_077_o0*`, the whale `_o0b`, and 0x8013B568..0x8013C98C — a target outside all four still needs checking - **same instruction multiset, one contiguous window, loads/registers ROTATED inside it** → pure sched1 statement-order (`gcc-2.7.2-map/sched.md` §S1/§S4). **Cheapest lever: brute-force it** — N independent statements, script all N! orders through `match_one` (24 runs ≈ 2 min) instead of reasoning about `rank_for_schedule` - **a two-constant `if/else` or `?:` result lands in **$v1** where the target reuses the condition's **$v0**** → **§76** — fold the condition into a NAMED local and overwrite that SAME variable with the two constants (an s16 temp, a fresh temp, an inner scope, and both ternary polarities all stay $v1; only reusing one `s32` local coalesces onto $v0). §76 reads as behemoth-only; it is not - **a load HOISTED above a store (match_one may call it a WIDTH class)** → usually **§76 regalloc**, not a width bug: a missing WAR dependency lets sched2 hoist it. Same fix — one variable for the compare temp and the result - **`ori $v0,0xffd8` in yours vs `addiu $v0,-0x28` in the target** → a NEGATIVE constant stored into an **unsigned** narrow local materialises via zero-extended `ori`; make the local **signed** to get `addiu`. It does NOT cost you the `lhu` on readback — gcc-2.7.2 emits `lhu` for any plain HImode load feeding an `sh` (P30 wave 4, byte-tested) - **LENGTH-DRIFT −1 and the missing instruction is a reg-to-reg COPY in a `jal` delay slot** → a narrow **prototyped** param (plain ANSI `s16 arg3`) — §43 is indexed as the K&R-definition lever, but the ANSI form is the fix as often - **your narrow load lost its load-delay `nop` right after an inline block move** → an `lwl/lwr`+`swl/swr` pair is a delay-slot **SPONGE** for the following load — the inverse of the §5a/§34 fence case. Align-1 4×u8 struct assign emits the inline form (§38 covers only the -O0 memcpy-call form) - **your `e = param_1` copy VANISHED (target addresses every field off a copy at a join block)** → cse.c `make_regs_eqv` keeps a param copy only when the new pseudo's live range escapes the cse block AND outlives the param's last mention — reach that by REUSING the same variable in a later block. Copy-in-the-arms gets hoisted into the delay slot; copy-assigned-in-an-arm goes global allocno → callee-saved + 2 prologue insns - **⚠️ the `.run/ghidra_c/.c` seed looks like a DIFFERENT function entirely** → it may be — overlays share VAs, so a seed can be decompiled from another overlay mapped at the same address. Trust the target `.s`, not the seed (P30 wave 4) ## By symptom ### delay slots & branches (66) - **§3-T4** — Branch polarity: invert the source condition to flip gcc's chosen branch L90 - **§5a** — Cross-jump tail-merge — gcc collapses two byte-identical blocks the original kept separate (FIX FOUND) L211 - **§3-D.** — THE CROSS-JUMP RATCHET (the sharpest new trap — `func_80131340` L-C) L3546 - **§3-C.** — Inner-block declaration does NOT delay slot allocation — BYTE-REFUTED L10106 - **§162** — CROSS-JUMP DIRECTION: the surviving copy is always the LATER one, so a BACKWARD `j` into a sibling arm is a source `goto` (P30 S48) L11304 - **§162** — The cross-jump "CALL veto" is a COUNT law, not a CALL law (BOUNDS §88a; P30 S48, `func_80189540`) L11340 - **§177** — 🔴 THE EPILOGUE RETURN-DELAY SLOT IS DECIDED BY YOUR SAVED-REGISTER SET, NOT BY SCHEDULING L17227 - **§176-A** — "SCHEDULE / DELAY-SLOT / LENGTH-DRIFT ±1" ⇒ check STATEMENT ORDER around the call first L17684 - **§186** — CROSS-JUMPING RUNS **AFTER** SCHEDULING, SO NO C-LEVEL BARRIER CAN STEER IT L18137 - **§194-H** — §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a branch fences the store that reads v, and when TWO independent stores compete for the one delay slot the fence picks the winner at ZERO length drift (a WIDTH/sh!=sw swap, not a nop). Corrects §167-42's reconstruction (pseudo arm → hard-reg arm) and supplies its first re-runnable A/B. L19356 - **§194-M** — A STORE in a CONDITIONAL branch's delay slot proves its C statement DOMINATES the branch — reorg can never pull a store out of either thread (gcc-2.7.2, -mips1) L19615 - **§195-D** — §195 — `masked_diff.mask_for` returns 0 for EVERY `j`/`jal` word, so an internal `j` destination is invisible to match_one, the permuter scorer AND every similarity tier: a control-flow semantic error (which calls execute) surfaces as a 1-instruction residual mislabelled `DELAY-SLOT / profile=schedule`, or as an outright false MATCH L19929 - **§195-E** — A 0/1 materialised at a JOIN immediately before the controlling `beqz`/`bnez` proves the source NAMED the condition — a truth expression in an `if`'s controlling position reaches `do_jump`, which has no value path L19997 - **§195-F** — fold-const.c:4825 canonicalises a `?:` whose THEN arm is zero (or constant against a non-constant ELSE) by inverting the condition and swapping the arms — so a ternary's written arm order is byte-inert there, and `c ? 0 : X` is unspellable: it always compiles as `!c ? X : 0` L20055 - **§195-G** — §NEW — TWO ARMS CALLING THE SAME CALLEE MERGE INTO ONE `jal` UNLESS EACH ARM'S OWN CODE CONSUMES THE RESULT: the C dial that keeps two call sites apart is WHERE the consumer test lives, and the branch SENSE you spell it with is byte-inert (jump.c:1737 canonicalises it) L20107 - **§195-J** — GTE / PsyQ op CALL-vs-INLINE is a PER-SITE SOURCE FACT, not a TU style — both forms coexist in one TU (measured in 2 TUs), and the tell is the target's own opcodes (`lwc2`/`sqr`/`swc2` vs `jal`), not the sibling. Costs -8 ins on func_8018505C. Corollary: the game's inline `sqr` macro emits TWO hazard nops, the SDK's `Square0` body emits ONE — so the inline form is provably not `Square0` inlined (a second instance of §187's SDK-vs-game GTE nop divergence). L20267 - **§195-N** — In a call-bearing chain of N≥2 `if (f(...)) return 1;` tests closed by `return 0;`, the LAST test must stay in STATEMENT form — the value form (`return f() != 0;` / `? 1 : 0` / `!!f()`) costs +1 `j` and empties the other N−1 delay slots. The cause is REORG block placement, not jump.c's `delete_jump`. L20464 - **§197-B** — A REPEATED COMPARE OF ONE VALUE AGAINST ONE CONSTANT IS DELETED BY cse's `qty_comparison_code` CHANNEL (the non-EQ complement of §165-03) — and a front-end-opaque mask on EITHER compare is a pure-C dial that keeps the target's second branch L20670 - **§199-F** — §164-36b — THE TARGET-HEAD FENCE IS A DELAY-SLOT THREAD SELECTOR, AND IT ONLY FIRES WHEN `mostly_true_jump > 0` (amendment to §164-36; its "−1 instruction" tell is falsified) L21030 - **§199-G** — At TWO case nodes the switch-vs-if oracle is not blind — but the tell is ALL tests positive + a trailing `j default`, NOT the first test's polarity L21084 - **§201-C** — §X — A CALL'S OWN DELAY SLOT AND THE UPSTREAM CONDITIONAL BRANCH'S SLOT COMPETE FOR ONE INSN (the call's argument copy), AND ONE STATEMENT'S POSITION RELATIVE TO THE CALL DECIDES BOTH — the residual is visible at the BRANCH, not at the call L21321 - **§204-A** — A COMPARE THAT APPEARS BOTH IN A BRANCH'S DELAY SLOT AND AGAIN ON THE FALL-THROUGH IS A JOIN WITH TWO INCOMING EDGES: THE TWO GUARDS ARE SEQUENTIAL `if`s, NEVER `if/else if` L21617 - **§211** — HOIST THE LOOP INIT ABOVE THE DOMINATING GUARD: it fills the guard's delay slot AND flips the counter/pointer register pair (P31 S58) L22425 - **§221** — A CONSTANT SHARED BY TWO STORES DIES AT THE CALL WHOSE DELAY SLOT REFILLS ITS REGISTER **(single observation — not yet cross-confirmed)** (P31 S58) L22829 - **§223** — READING A `jal` DELAY SLOT: the value in it was produced BEFORE the call, so it is NEVER that call's return (P31 S58) L22887 - **§224** — CROSS-JUMP: WRITE THE DUPLICATE, AND READ A SHARED DELAY SLOT AS THE MERGE SIGNATURE (P31 S58) L22949 - **§232** — WHEN THE `jr` DELAY SLOT'S STORE STORES THE RETURN VALUE, TIE THEM WITH ONE PSEUDO **(single observation — not yet cross-confirmed)** (P31 S58) L23231 - **§243** — SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P31 S58b) L23731 - **§247** — TWO BRANCHES TO **ONE** LABEL MEANS THE SOURCE CONDITION IS NEGATED (P31 S58b) L23883 - **§224** — addendum (P31 S58b) — CROSS-JUMP MERGES *CALLS*, AND THE DELAY SLOT IS THE DISCRIMINATOR L24570 - **§263** — A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCHEDULE (P31 S59, byte-proven) L24937 - **ADD-4** — → §225-3/-4 addendum — THE MIRROR ROW: VALUE-RETURN IN THE *TAKEN* ARM + TRAILING BARE `return 0` L25169 - **ADD-5** — → §1/I1 addendum — THE INVERTED RANGE TEST: `(u32)(x-lo) >= N` WITH THE ZERO-ARM AS THE TRAILING `else` L25183 - **ADD-5** — → §267-ADD-6 / §172b-1 addendum — THE SIGN-TEST FACE: `(s16)v < 0` IS `sll 16` + `bgez` ON THE SHIFTED COPY, RAW STAYS LIVE — AND THE HALFWORD-ABS SHAPE NEEDS NO RITUAL L25421 - **ADDENDUM** — to §224 — CROSS-JUMP: THE DUPLICATE CAN BE A PLAIN STORE, NOT ONLY A CALL L26014 - **ADDENDUM** — to §172a — RE-READING MEMORY (NOT NAMING A TEMP) IS WHAT KEEPS AN INCREMENT'S DELAY-SLOT FILL ALIVE L26314 - **ADDENDUM** — to §225 / §256 — A GOTO TO A SHARED SET-POINT PREVENTS IF-CONVERSION FROM COLLAPSING A LATER BRANCH TEST L26345 - **§277** — RETURN-TAIL C SPELLING PICKS THE DELAY-SLOT-FILL vs TRAILING-MOVE TOPOLOGY, AND A NARROWER SECOND VARIABLE KEEPS TWO PSEUDOS INSTEAD OF ONE (P31 S60; `func_801846F0` ov_SC03_104, `func_801A44C4` md_SC07_004, both byte-proven) L26571 - **ADDENDUM** — to §22 (`volatile`-qualified-global reload lever, cookbook ~L1922) — for a NON-constant, same-address double RMW, a plain memory clobber beats `volatile`, and `volatile` actively breaks a delay-slot fill L27236 - **ADDENDUM** — to §220 — REFERENCING THE RAW PARAMETER (NO NAMED COPY, NOT EVEN A PIN) LETS THE CALLEE-SAVED SPILL LAND IN THE FIRST CALL'S OWN DELAY SLOT L27490 - **ADDENDUM** — to §252 — a `>=0`/`<0` split on an unconditionally-decremented value needs the POSTFIX operator INSIDE the branch condition, not a prior statement L27548 - **ADDENDUM** — to §20's cross-jump EXPLOIT bullet (func_8017E360, ov_SC05_007 — wave dg) L27842 - **ADDENDUM** — §NNN — sharpens §55a / §164-37 / §165-28 (switch-vs-tree cluster): A SHAPE THAT LOOKS LIKE A SWITCH DISPATCH CAN BE PLAIN NESTED `if`s WHOSE SHARED BODY WAS TRIPLICATED BY THE SOURCE AND THEN CROSS-JUMP-MERGED BACK DOWN L28233 - **§285** — PRE-INITIALIZING A VARIABLE WITH A SHARED CONSTANT BEFORE A BRANCH MAKES BOTH ARMS OF THE FOLLOWING if/else DESTRUCTIVELY REUSE THE SAME DESTINATION REGISTER FOR THEIR BITWISE RESULT (byte-proven; `func_801811F0`, ov_SC03_102, independently rediscovered across waves di/dj) (P31 S60; waves #, byte-proven) L29088 - **§286** — FOLD A STATEMENT'S SIDE EFFECT INTO A COMMA-EXPRESSION IN AN ARGUMENT POSITION TO PLACE ITS RTL RELATIVE TO A CALL'S OWN DELAY SLOT (P31 S60/dj; `func_80180A88`, ov_SC06_010, byte-proven 411/411) (P31 S60; waves #, byte-proven) L29092 - **§291** — THE DELAY-SLOT FALSE-VALUE: A CONDITIONAL BRANCH'S ZERO ARM MUST BE A FALL-THROUGH-ADJACENT BLOCK ENDING IN AN EXPLICIT JUMP, OR REORG CANNOT MATERIALIZE IT INSIDE THE BRANCH'S OWN DELAY SLOT (P31 S60; waves #, byte-proven) L29112 - **ADDENDUM** — to §253 — SECOND byte-proven card (upgrade from single-observation), and the placement face: postfix-in-condition parks the RMW store in the branch delay slot L29346 - **§306** — A HAZARD `nop` IN FRONT OF A DIV-RESULT STORE IS A STATEMENT-ORDER DEFECT: THE INDEPENDENT TRAILING STATEMENT MUST BE WRITTEN *BEFORE* THE DIVISION-CONSUMING ONE (P31 S62 T4; byte-proven func_8017E7D0) L29884 - **§308** — A NAIVE FINAL `goto` DELETES THE TARGET'S `j`+`nop`: SPELL THAT EDGE AS A TAUTOLOGICAL `if` SO cse FOLDS IT TO AN UNCONDITIONAL JUMP *AFTER* jump1's WINDOW HAS CLOSED (P31 S63 t5e-t5i; byte-proven func_80180808) L29993 - **§308a** — A TWO-INSTRUCTION ARM STUB LAID *BEFORE* THE OTHER ARM AND REACHED ONLY BY A TRAILING `j` SURVIVES ONLY BEHIND A SHORT-CIRCUIT `&&` GUARD PLUS A SEPARATE RE-TEST; EVERY PLAIN if/else-if, GOTO-LADDER, NESTED-INVERTED-if AND 1-/2-NODE switch LETS jump1 INLINE IT (−2 ins) (P31 S63 t5e-t5i; byte-proven func_80180808) L30022 - **§309** — A FRAME-ADDRESS EQUIVALENCE RIDES *THROUGH* A SKIPPABLE CONDITIONAL: POST-`if` STORES FOLD ONTO `$sp` UNLESS THE `if` SURVIVES jump1 AS A BARRIER-PRECEDED DIAMOND (P31 S63 t5e-t5i; byte-proven func_801812AC) L30056 - **§311** — A JOIN-BLOCK STORE CAN SWALLOW AN ARM'S LOAD-DELAY `nop`: WHEN ONE ARM'S VALUE COMES FROM A LOAD, DUPLICATE THE STORE INTO BOTH ARMS (P31 S64 t5j-t5m; byte-proven func_8017D7E0) L30127 - **§314** — AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES ALL COMPILE BYTE-IDENTICAL, AND NONE REACHES A TARGET THAT SINKS THE NEGATIVE ARM INTO AN OUT-OF-LINE BLOCK (P31 S65; UNPROVEN, gate-refused draft func_80181720) L30251 - **Addendum** — A §5a BARRIER SITS IN REORG'S FILL WINDOW TOO, SO ITS SIDE OF THE INNER `goto` IS CHOSEN P (func_80181D08) L30499 - **§NNN** — REFUTED: A `sw %lo(SYM)($at)` IN A JUMP'S DELAY SLOT WITH ITS `lui $at` STRANDED ABOVE THE JUMP IS **ASSEMBLER** MACRO-EXPANSION, NOT A cse SPLIT — OUR PIPELINE CANNOT EMIT IT FROM ANY C, AND THE +1 IS A `maspsx` GAP (P31 S66 round2, `func_8005DBD8`, byte-probed) L30601 - **Addendum** — REGALLOC-PERM store in a cross-jump shared tail defeats §137's barrier (func_8017EDE8) L30640 - **Addendum** — LENGTH-DRIFT nop clears when offset math precedes the symbol launder (func_80039B20) L30644 - **Addendum** — Return-0 statement order: extra j+move vs delay-slot fusion (func_8018BB50) L30731 - **§331** — OPEN GAP: NO LEVER ELIMINATES AN *UNWANTED DUPLICATE* COPY AT A BRANCH-TARGET BLOCK HEAD (P31 S67; main/func_80013154, closeness 12, NOT solved) L31069 - **§332** — THE maspsx `la`-IN-A-DELAY-SLOT GAP: gcc EMITS A SYMBOLIC ADDRESS LOAD AS ONE ATOMIC length-2 INSN, SO IT CAN NEVER FILL A JUMP DELAY SLOT — 6 FUNCTIONS FLEET-WIDE, NONE BANKABLE FROM C (P31 S67; byte-traced main/func_80062144, func_8005DBD8) L31081 - **§336** — THE §5a CROSS-JUMP BARRIER GOES AT THE *BOTTOM* OF THE TWIN, NOT THE TOP: `find_cross_jump` WALKS BACKWARD FROM THE CONVERGING JUMP (P31 S67; byte-proven ov_SC05_001/func_80183C9C) L31159 - **§339** — A 2-CASE SWITCH OMITS THE LOW-BOUND RANGE TEST, SO THE PRESENCE OF `slti/bnez` BETWEEN THE `beq`s IS A **COUNT TELL** FOR A THIRD CASE NODE (P31 S67; byte-proven ov_SC02_005/func_80190538, 197 ins) L31191 - **§346** — `c ? X : -X` TAKES expand_expr's COND_EXPR **SINGLETON** PATH (copy, then negate IN PLACE) — AN if/else STATEMENT GIVES THE TWO-ARM FORM (P31 S67; byte-proven ov_SC03_102/func_80180C38, closed the last instruction) L31332 - **§370** — ★★ — A **HARD BOUND** FROM sched.c, AND THE reorg SLOT-STEAL DIAGNOSTIC (P31 S68; main/func_8001BC6C, 69 ins, NOT closed — 33 → 28 over ~45 measured compiles) L31848 - **§396g** — ★★ — A GUARD LADDER'S RUNGS MUST STAY SYMMETRIC OR `reorg.c` LOSES ITS BRANCH REDIRECT (P31 S69; byte-proven ov_SC03_028/func_80184C90, 92 ins) L32828 - **§428** — ★★★ — A ZERO-BYTE CROSS-JUMP BARRIER: ADVANCE THE POINTER INSIDE EACH SWITCH ARM (P31 S72; `main/func_80026D64`, 218 ins, MATCH in 2 compiles) L33964 ### instruction scheduling (100) - **§3-T2** — Source statement order drives instruction scheduling L78 - **§3** — When a diff is pure scheduling → decomp-permuter (harness built, Phase 6) L107 - **§10** — Closing the regalloc/scheduling hard tail by hand (LZSS, Phase 7 session F — the full close) L854 - **§25** — The "schedule" class is mostly COALESCING (pin-crackable), not scheduling; + the gate two-stage + h_exact over-counts ×134 (Phase 21, cont.6) L2155 - **§3-The** — crack: the residual was gcc-2.7.2 COPY-COALESCING, not the instruction scheduler L2167 - **§3-The** — genuine scheduler — `rank_for_schedule` (sched.c), for when it IS scheduling L2183 - **§3-The** — genuine schedule WALLS (do NOT re-grind — stub) L2192 - **§32** — The region-a CAMERA-GIANT idiom set: struct-base hoisting + 4 sibling levers (Phase 24 T7, Fable5-cracked on `func_80129CF8` 191 ins, match_one MATCH; transferable to the 6 sibling giants) L2445 - **§34** — The `func_80138ED0` giant crack: gcc-2.7.2's **3-qty sort bug** + the **zero-byte asm allocation toolkit** + the **giv-init fence** (Phase 24 T5; Opus→close=21, Fable5→MATCH ×134) L2477 - **§35** — The region-a sibling-giant harvest: difficulty ≠ $s-reg count (it's global-array hoisting) + the banking recipe + new loop idioms (Phase 24 T7, 5 parallel Opus-Max agents) L2494 - **§3-A4** — SINK THE CONSUMER CALL INTO THE ARMS (the inverse of A1; `func_8016AB6C`, byte-proven) L3482 - **§49** — The LUID DIAL: a zero-byte SCHEDULING dial (the sched.c analogue of §47) — `func_8017A4AC` (536 ins ×134), Phase 26 session 8 L3565 - **§66d-5** — `residual_class`'s "structural ⇒ permuter CPU is waste" is WRONG for schedule permutations (measured, Phase 29 SESSION-18) L5529 - **§3-The** — attribution primitive (use this before calling anything a scheduling residual) L6095 - **§3-The** — scheduling rules (refining §135-2 and §135-4) L8950 - **Consequence** — for the family (a real scheduling decision) L10133 - **§148** — The loop.c hoisting THRESHOLD is arithmetic you can compute, and the `?:` clamp that folds to MIN_EXPR (P30 S42, `func_8017C6F4`, 947 ins) L10183 - **§3-A.** — `move_movables` hoists iff `threshold × savings × lifetime ≥ insn_count` — and you can read it L10189 - **§16Z** — SHARPENS *(sharpens §37 "the /s-DEP LATTICE", §136-13, §136-14, §16Xy, §162q, `gcc-2.7.2-map/sched.md` §64)* L14398 - **§175** — A CALLER-SAVED REGISTER PIN CAN BE A CORRECTNESS BUG, NOT JUST A SCHEDULING CHOICE (P31 wave H, 2026-08-15) L16780 - **§177** — 🔴 THE EPILOGUE RETURN-DELAY SLOT IS DECIDED BY YOUR SAVED-REGISTER SET, NOT BY SCHEDULING L17227 - **§3-B.** — A `return ` IS A PRIORITY-1 HARD-REG SET THE SCHEDULER PLACES FIRST (func_8001BE30, 92 ins) L17312 - **§176-A** — "SCHEDULE / DELAY-SLOT / LENGTH-DRIFT ±1" ⇒ check STATEMENT ORDER around the call first L17684 - **§176-C** — 🔴 WALL REFUTATION: a hard-register pin CANNOT schedule around a call, because of a genuine gcc-2.7.2 bug L17754 - **§186** — CROSS-JUMPING RUNS **AFTER** SCHEDULING, SO NO C-LEVEL BARRIER CAN STEER IT L18137 - **§193-C** — gcc-2.7.2 cross_jump merges the SCHEDULED common SUFFIX only — there is no prefix/head merge, so §8/§48-A1's "duplicate into both arms and cross_jump refunds it" is a TAIL-only lever L18592 - **§193-D** — A COMPILER-GENERATED ARGUMENT COPY IS A `optimize_reg_copy_1` TRIGGER: when the pointer's LAST use in a block precedes a call that takes the pointer *as a register*, sched1 hoists the implicit `move $aN,$sN` to the block top and local-alloc re-bases the WHOLE block's memory operands onto `$aN`. The only C dial is a label (the §165-24 goto-join) between the block and the call. L18626 - **§193-F** — §148-A2 — The `threshold -= 3` STAIRCASE: `move_movables` admits a COUNT of invariants, not a boolean — two identical merged constants split hoisted/not-hoisted by LIST ORDER, and `insn_count` picks the rank cutoff L18727 - **§194-A** — A zero-byte scheduling fence goes AFTER the defining statement to make that computation emit FIRST in its block — and the barrier predicate is `volatile`-or-colon-less, not the `"memory"` clobber (COMPLEMENTS §165-40; does not refute it) L18995 - **§194-D** — Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication + attribution correction of §165-17, NOT a new argument-register law L19162 - **§194-H** — §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a branch fences the store that reads v, and when TWO independent stores compete for the one delay slot the fence picks the winner at ZERO length drift (a WIDTH/sh!=sw swap, not a nop). Corrects §167-42's reconstruction (pseudo arm → hard-reg arm) and supplies its first re-runnable A/B. L19356 - **§194-K** — Blind sched1's alias oracle with a second SET of a pointer pseudo — the first zero-byte, non-volatile, dependence-CREATING lever (corrects §167-05's "volatile is the only door"; fourth consumer of reg_n_sets) L19489 - **§194-N** — §193-D's C dial is misstated: the lever is a SURVIVING CODE_LABEL (a label with a real incoming edge), not "a label between the block and the call" — a bare label, or a `goto L; L:` pair whose target is the next active insn, is deleted by jump1 (jump.c:663-669 → delete_insn → jump.c:3458-3461, and jump.c:243 for the bare case) long before sched1/local-alloc, and costs exactly zero bytes L19673 - **§195-D** — §195 — `masked_diff.mask_for` returns 0 for EVERY `j`/`jal` word, so an internal `j` destination is invisible to match_one, the permuter scorer AND every similarity tier: a control-flow semantic error (which calls execute) surfaces as a 1-instruction residual mislabelled `DELAY-SLOT / profile=schedule`, or as an outright false MATCH L19929 - **§199-A** — §189-A's asm→source inference is byte-FALSE: an interloper between a split constant's `lui`/`ori` is a SCHEDULE fact, not a source fact — and the separator is the BIRTHING BOOST, not a "priority floor" (§189-A's split-timing half survives; its "no statement order / no pin" absolute and the candidate's own forward-scheduler narrative both fall) L20769 - **§199-F** — §164-36b — THE TARGET-HEAD FENCE IS A DELAY-SLOT THREAD SELECTOR, AND IT ONLY FIRES WHEN `mostly_true_jump > 0` (amendment to §164-36; its "−1 instruction" tell is falsified) L21030 - **§205** — THE CHAINED ASSIGNMENT IS ITS OWN SCHEDULING DIAL: `*b = *a = v;` moves an argument copy that no local, no pin and no statement reorder will move (P31 S56) L22108 - **§211** — HOIST THE LOOP INIT ABOVE THE DOMINATING GUARD: it fills the guard's delay slot AND flips the counter/pointer register pair (P31 S58) L22425 - **§219** — COMPOUND `+=`, FULL ASSIGNMENT, AND AN EXPLICIT TEMP ARE THREE DIFFERENT SCHEDULES OF ONE READ-MODIFY-WRITE (P31 S58) L22762 - **§30** — addendum (P31 S58) — the `/s` grant closes a SCHEDULING residual and a REGISTER residual with one edit L23267 - **§165-40** — addendum (P31 S58) — the barrier goes at the COPY SITE, not inside the region it protects L23299 - **§239** — TWO-STATEMENT INTEGER-SPACE MATERIALISATION REORDERS `la` vs `sll`; AND THE PLUS-TREE OPERAND ORDER (P31 S58b) L23611 - **§243** — SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P31 S58b) L23731 - **§245** — THE CALL'S ARGUMENT LIST IS A SCHEDULING SLOT (P31 S58b) L23796 - **§211** — addendum (P31 S58b) — INIT PLACEMENT: FIVE MORE DIALS BEYOND THE GUARD HOIST L24326 - **§263** — A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCHEDULE (P31 S59, byte-proven) L24937 - **ADD-8** — → §245 addendum — THE RE-TIE BARRIER PINS CALL-ARG SETUP TO SOURCE ORDER (construct 6) L25213 - **ADDENDUM** — to §244 — THE COMPILER BARRIER, NOT `volatile`, BLOCKS A STORE→LOAD HOIST THROUGH A POINTER-OFFSET FIELD L26079 - **§281** — GROUP COPY-THEN-RMW BY OPERATION KIND, NOT FIELD BY FIELD: sched1 does the interleaving, the source must not (P31 S60; wave cf, func_80180FB4, ov_SC03_111, byte-proven) L27000 - **§NNN** — A DEPENDENT COPY-THEN-RMW BLOCK MUST BE SOURCE-GROUPED BY OPERATION KIND, NOT BY FIELD, AND SCHED1 DOES THE INTERLEAVING (P31 S60; `func_80180FB4`, ov_SC03_111, byte-proven) L27001 - **§282** — gcc's OWN LOOP REVERSAL PUTS THE COUNTER INIT AFTER THE HOISTED MOVABLES — a position no hand-written down-count can reach (P31 S60; wave cf, func_80180DD8, ov_SC04_005, byte-proven) L27046 - **§NNN** — WRITE THE UP-COUNT LOOP: gcc's OWN REVERSAL PRODUCES A COUNTER-INIT INSTRUCTION THAT LANDS AFTER HOISTED MOVABLES, WHICH A HAND-WRITTEN DOWN-COUNT LOOP CANNOT REPRODUCE (P31 S60; `func_80180DD8`, ov_SC04_005, byte-proven) L27047 - **ADDENDUM** — to §195-N — a GNU statement-expression slider must sit INSIDE the conditional arm's value position, not as a post-hoc barrier, to block the store-flag transform on a ternary chain L27317 - **ADDENDUM** — §NNN — sharpens §167-13's boundary: CHAINING TWO IDENTICAL SIDE-BY-SIDE STORES INTO ONE C ASSIGNMENT STATEMENT IS A MID-BLOCK SCHEDULING-PRIORITY DIAL, NOT ONLY A STORE-ORDER SPELLING L28270 - **§284** — COMBINE CAN REASSOCIATE TWO SEQUENTIAL BITWISE-AND MASKS INTO ONE AGAINST THE PRE-MASK VALUE; AN ASM IN/OUT FENCE RIGHT AFTER THE FIRST MASK BLOCKS IT (P31, wave dd, `func_8018087C`, ov_SC04_020, byte-proven) (P31 S60; waves #, byte-proven) L29084 - **§298** — THE SHARED-TAIL POSITION DIAGNOSTIC: A FOLDED TAIL *BETWEEN* SWITCH ARMS MEANS PER-ARM DUPLICATED STATEMENTS, NOT A HOISTED POST-SWITCH STATEMENT (P31 S61; waves g0e/g0f, `func_80181B68`, ov_SC06_016, byte-proven 68/68) L29552 - **§299** — TWO INDEPENDENT EXTRACTION CHAINS EMIT CONTIGUOUSLY INSIDE ONE EXPRESSION; ONLY A STATEMENT BOUNDARY MAKES THE SCHEDULER INTERLEAVE THEM (P31 S61; wave m0a, `func_8003A404`, main, byte-proven 8/8) L29577 - **§307** — THE BRUTE-FORCE-THE-STATEMENT-ORDERS LEVER HAS A BOUND: A FAN-OUT COPY'S PRIORITY IS SCHEDULER-INTERNAL, AND NO SOURCE ORDER REACHES IT (P31 S63; byte-evidenced NEGATIVE result, main wave) L29949 - **§309** — A FRAME-ADDRESS EQUIVALENCE RIDES *THROUGH* A SKIPPABLE CONDITIONAL: POST-`if` STORES FOLD ONTO `$sp` UNLESS THE `if` SURVIVES jump1 AS A BARRIER-PRECEDED DIAMOND (P31 S63 t5e-t5i; byte-proven func_801812AC) L30056 - **§314** — AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES ALL COMPILE BYTE-IDENTICAL, AND NONE REACHES A TARGET THAT SINKS THE NEGATIVE ARM INTO AN OUT-OF-LINE BLOCK (P31 S65; UNPROVEN, gate-refused draft func_80181720) L30251 - **Addendum** — A §5a BARRIER SITS IN REORG'S FILL WINDOW TOO, SO ITS SIDE OF THE INNER `goto` IS CHOSEN P (func_80181D08) L30499 - **Addendum** — Loop sentinel wrongly hoisted out of a call-containing loop (func_801835B0) L30503 - **Addendum** — REGALLOC-PERM: one memory-clobber barrier insufficient — stack two (func_80182D1C) L30533 - **Addendum** — A plain global-RMW triplet is a fillable scheduling window too — zero drift, and the fille (func_800CB058) L30543 - **Addendum** — Shared arithmetic that CONSUMES the selected value cannot be hoisted above the `if` — narr (func_80183DA0) L30547 - **Addendum** — Truncating assign must be the lazy `||` operand, not hoisted (func_80012B58) L30570 - **Addendum** — WIDTH sh!=lhu residual: hoist a sibling field's load first (func_80182538) L30582 - **Addendum** — REGALLOC-PERM store in a cross-jump shared tail defeats §137's barrier (func_8017EDE8) L30640 - **Addendum** — Volatile zero-byte asm slider scrambles the prologue schedule (func_801814B0) L30757 - **Addendum** — §194-A addendum — the bare/colon-less fence measured NULL and only the "memory"-clobber fo (func_8017E464) L30771 - **§336** — THE §5a CROSS-JUMP BARRIER GOES AT THE *BOTTOM* OF THE TWIN, NOT THE TOP: `find_cross_jump` WALKS BACKWARD FROM THE CONVERGING JUMP (P31 S67; byte-proven ov_SC05_001/func_80183C9C) L31159 - **§340** — §194-K COROLLARY: **FLIP THE FALSE EDGE YOU CANNOT DELETE.** A "scheduler" residual can be sched.c's ALIAS ORACLE emitting a FALSE true-dependence; source order chooses its DIRECTION (P31 S67; byte-proven ov_SC03_107/func_8017CF48, 10 -> 0 in one compile, zero bytes) L31219 - **§341** — AN HImode STORE TEMP REWEIGHTS A sched2 TIE-BREAK THAT NO STATEMENT ORDER CAN REACH (P31 S67; byte-proven ov_SC03_006/func_801823B8, last 4 ins) L31253 - **§349** — RE-ASSIGN A BASE POINTER AT THE END OF THE LOOP BODY TO MAKE `n_times_set > 1` — THAT KILLS BOTH loop.c's INVARIANT HOIST **AND** THE ADDRESS GIV, REPRODUCING A REMATERIALISED `addiu $aN,$sp,K` (P31 S67; byte-proven ov_SC06_029/func_801804C8, 255 ins) L31424 - **§350** — A ZERO-BYTE RE-TIE SETS `reg_n_sets=2`, WHICH KILLS sched1's `birthing_insn_p` LAUNCH_PRIORITY BOOST — THE MECHANISM BEHIND "MY ADDS ARE GLUED TO THEIR STORES" (P31 S67; byte-proven ov_SC04_011/func_80180B24, 215 ins) L31445 - **§352** — ⚠ TWO IDENTICAL `__asm__ __volatile__("")` BARRIERS **MERGE WITH EACH OTHER** — SPELL THE SECOND ONE DIFFERENTLY (P31 S67; byte-proven resident/func_800D128C, measured closeness 105 when they merged) L31496 - **§361** — ★ — A LOOP-TAIL BYTE SIGNATURE THAT NAMES ITS SOURCE SHAPE — AND THE LAW THAT A "SCHEDULING TIE" MAY BE YOUR OWN EARLIER LEVER (P31 S68; byte-proven main/func_800241C0, fable escalation, 19 → 0 in 3 iterations) L31647 - **§370** — ★★ — A **HARD BOUND** FROM sched.c, AND THE reorg SLOT-STEAL DIAGNOSTIC (P31 S68; main/func_8001BC6C, 69 ins, NOT closed — 33 → 28 over ~45 measured compiles) L31848 - **§373** — ★★★ — THE **DEAD-RESET CSE-BREAKER**, THE PIN THAT BREAKS A sched2 ANTI-DEP, AND WHY AN `asm` CAN NEVER RAISE PRIORITY (P31 S68; byte-proven ov_SC06_010/func_8017E764, 438 ins, fable escalation closed 8 → 0) L31987 - **§3-2.** — A REGISTER PIN THAT DELETES A sched2 ANTI-DEPENDENCE L32007 - **§3-3.** — HARD FACT FOR THE SCHEDULING MAP — an `asm` ALWAYS has priority 1 L32020 - **§379** — ★★★ — **MEM_IN_STRUCT_P**: THE SAME LOAD, WRITTEN AS A STRUCT MEMBER, SCHEDULES WHERE A CAST CANNOT (P31 S69; byte-proven main/func_80021284 220 ins and main/func_8002D904 217 ins, found INDEPENDENTLY by two agents) L32174 - **§381** — THE `insn_count` HOIST THRESHOLD IS A DIAL YOU CAN READ WITH `cc1 -dL` (P31 S69; four independent uses in one wave) L32226 - **§385** — ★★★ — THE **SCHED2 PRIORITY-DONOR ASM**: closing the "hoisted-invariant vs IV-init preheader swap" class (P31 S69; byte-proven main/func_80038A58, 347 ins, fable escalation 2 → 0) L32319 - **§393** — ★★ — THE **BIRTHING BOOST**: A SINGLE-SET LOCAL IS SCHEDULED LAST; GIVE IT A SECOND SET (P31 S69; byte-proven ov_SC02_017/func_8017FCFC) L32553 - **§3-B.** — THE SCHEDULER DIALS (the dominant residual family this wave) L33118 - **§424** — ★★★ — EQUAL-PRIORITY STORES COME OUT **REVERSED**: sched1's LUID tie picks the LAST statement first (P31 S71; byte-proven `ov_SC07_006/func_801890FC`, 387 ins) L33817 - **§428** — ★★★ — A ZERO-BYTE CROSS-JUMP BARRIER: ADVANCE THE POINTER INSIDE EACH SWITCH ARM (P31 S72; `main/func_80026D64`, 218 ins, MATCH in 2 compiles) L33964 - **§430** — ★★★ — A GOTO INTO A LOOP IS FINE; HAND-HOIST THE CONSTANTS IT COSTS YOU (P31 S73 — **this section previously said the OPPOSITE and was wrong; the refutation is kept below**) L34083 - **§3-A.** — `reg_n_sets` IS A ONE-LINE SCHEDULING DIAL — `t = t + 1;` vs `*p = t + 1` (CONFIRMED: `ov_SC03_105:func_80180ABC`, 257 ins, banked from closeness 250) L34956 - **CORRECTION** — to §182/§188 (P31 S76) — THE EPILOGUE "WALL" IS AN ORACLE ARTIFACT IN THE REORDER ISLAND L35192 - **§460** — `-dS` PRINTS THE SCHEDULER'S READY LIST WITH PRIORITIES. STOP INFERRING IT FROM RTL ORDER. L35232 - **§465** — THE ASPSX SLOT-HOP: A GAP OUR REORDER SUBSTITUTE CANNOT CLOSE (`main:func_8005F830`) L35294 - **§464** — FOUR VOLATILE/BARRIER LEVERS FROM `main:func_8005DE78` (141 ins → MATCH) L35338 - **§470** — FOUR CSE/SCHED LEVERS FROM `main:func_800301C8` (170 ins, 133 → 18) L35447 - **§472** — 🔴 §148-A's HOIST THRESHOLD IS 29, NOT 58, WHEN THE LOOP CONTAINS A CALL L35499 - **§475** — THE `"memory"` FENCE AS A *CSE INVALIDATOR*, AND `(b*3)<<3` INSTEAD OF `b*24` L35591 - **§476** — 🔴 A HARD-REGISTER PIN DESTROYS TWO THINGS COMBINE AND SCHED1 NEED (`func_800226C0`, 670 ins → MATCH) L35622 - **§482** — ★★★ — TWO INDEPENDENT RE-TIES, ORDERED: WHEN ONE BARRIER FIXES ONE RESIDUAL AND CREATES THE OTHER (P31 S77, `main:func_8006252C`, 30 ins → MATCH) L35942 - **§456** — The lever-removal move catalog: what the S104 agents found under ~205 pins, barriers and launders (Phase 36 T7) L37765 ### register allocation & pins (140) - **§10** — Closing the regalloc/scheduling hard tail by hand (LZSS, Phase 7 session F — the full close) L854 - **Residual** — A — commutative `|`/`&`/`+` result lands in the wrong source-operand register L875 - **Register-allocation** — ORDER (call-crossing $s0/$s1 swap) → FORCE it with register pins (byte-proven) L1407 - **§3-The** — LOOSE-TYPING wall is real for narrow params (Phase 16, reconfirmed) L1451 - **§3-THE** — CAP: at this tail the match_one→gate gap is the LOOSE-TYPING CALL-GRAPH wall (not cheap plumbing) L1660 - **§22** — DEF-side loose-typing recovery + grinder blacklist (Phase 21) L2028 - **§24** — The `ov_SC01_077_a.c` split-file vein: split-aware propagation, but loose-typing-gated (Phase 21) L2108 - **§25** — The "schedule" class is mostly COALESCING (pin-crackable), not scheduling; + the gate two-stage + h_exact over-counts ×134 (Phase 21, cont.6) L2155 - **§42** — The F-band ≤28 regalloc crack wave: register-pin/DENSITY levers beat the permuter (Phase 25 T7 F-band, 2026-07-10; Ultracode 9-worker wave, 4/9 banked byte-identical, 266 swept ×134) L2938 - **§43** — The K&R s16-param definition DISSOLVES the "narrow-param wall" for by-value register args (Phase 25 task A, Fable5 crack of the 369-ins giant `func_80166994` ×134, 2026-07-11) L3233 - **§45** — The flagship `func_80133CD4` crack (399 ins ×134): the merged-variable permutation-breaker + the 1-death local-alloc gate (Phase 25 task A giant escalation, Fable5 gdb-on-cc1, 2026-07-11) L3336 - **§47** — The live-length SLIDER: splitting a global.c allocno-priority TIE with one zero-byte asm (Phase 26 session 8, Fable5 Max, byte-proven on `func_8017BEBC` 952 ins ×113) L3403 - **§48** — The 12-core jr crack wave: the ALLOCNO-PRICING dials and the EBB rule (Phase 26 session 8, Ultracode, 9/12 MATCH first pass) L3443 - **§3-A.** — ALLOCNO-PRICING DIALS — move a value into the register you want, byte-neutrally L3451 - **§52** — The WALKER-FAMILY skeleton: 6 regalloc-order levers + a deeply-characterized intrinsic wall (`func_80178004`, 165 ins ×134; Phase 26, Fable5, 2026-07-15) L3951 - **§52a** — The regalloc sibling wave: new levers + two new wall classes (cheap-Opus applying §52, 2026-07-15) L3997 - **§52b** — Sibling wave 2: more de-pin levers, a third wall class, and the match_one→whole-binary gap at scale (2026-07-15) L4035 - **§66d-1** — What transfers between giants is the LOOP, not the PIN L5318 - **§70** — The giv-init base register: walk the PARAMETER, not a copy of it (Phase 29 SESSION-18, `func_801777BC`) L5657 - **§71** — Before mapping a giant, look for an already-matched SIBLING beside it (Phase 29 SESSION-18, `func_8017D960`, 3,338 ins) L5698 - **§72** — A `register __asm__` pin is a PREFERENCE, not a reservation (Phase 29 SESSION-18, `func_8017F510`) L5764 - **§74** — Auditing a pinned draft: the §72 hazard is CALLER-SAVED pins spanning a call, and only the disassembly can tell you (Phase 29 SESSION-19, `func_8017D960` b2, 5 pins) L5860 - **§76** — The allocno CLASS (local vs global) is the dominant regalloc lever, and C reaches it ONLY through declaration scope and variable reuse (Phase 29 SESSION-19, behemoth #3 `func_8017F510` 1,511 ins, 97 → MATCH, pin-free) L6066 - **§78** — A LENGTH drift can be a register grant in disguise; and `fold` never leaves a literal first in an `|` chain (Phase 29 SESSION-19, behemoth #2 `func_8017D960` 3,338 ins, 1806 → 0, pin-free) L6212 - **§76** — confirmed at scale, and a pin nuance L6289 - **§80** — A do-not-re-buy entry is scoped to its BASE, not to the function; and the pin's hidden cost is an unconditional `qty_phys_sugg` (Phase 29 SESSION-19, `func_8017BF14` 45 → 0) L6308 - **§3-The** — pin's hidden cost, with the citation L6329 - **§83** — The parameterised-repeat law, the spill-area trap, and why a per-case edit cannot move a per-case symptom (Phase 29 SESSION-20, `func_80183814` 5,122 ins, cold-ish start → 36 structural / 99.3%) L6450 - **§83c** — TRAP: a "dead local" in a prior draft may be gcc's OWN spill area L6483 - **§86** — Pinned-exemplar templatability is a PER-FAMILY property, not a per-member rate; and the §42e pin guard is now over-conservative (Phase 29 SESSION-20) L6627 - **§3-Two** — further notes worth keeping L8286 - **§137** — REGALLOC-PERM is a TWO-COMPILE ARITHMETIC PROBLEM, not a permuter job L9382 - **§3-The** — same swallow, twice more, in the integration spine L9747 - **§3-B.** — A `?:` on MEMORY operands costs ~16 bytes of invisible frame; on REGISTER operands, zero L10095 - **§3-D.** — A lone `$t8`/`$t9` in the target is RELOAD SCRATCH — reproduce the spill, don't pin the register L10111 - **§3-C.** — A zero-byte ALLOCNO-PRIORITY slider L10222 - **§150** — A register ROTATION across symmetric blocks is VARIABLE-IDENTITY evidence, not an allocator tie (P30 S43, `func_8017C6F4`, 947 ins ×4) L10350 - **§152** — BYTE SIZE is the family key that name- and h_seq-grouping both miss (P30 S43, the 0xECC family: 1 crack → 12 overlays → 11,364 ins) L10456 - **§155** — hi/lo literal scanning MUST track base registers (S45) L10596 - **§158** — The RANGE-EXTENDER: a fourth zero-emission asm lever completes the allocno toolkit (P30 S46 tier-3, `func_8017CE58`, 733 ins) L10750 - **Bonus** — facts worth keeping L10795 - **§156** — THE PREFERENCE-DONOR MERGE: cross-region variable reuse is what fills a0-a3, and a call-arg use in ONE region steers the fill in ALL of them (P30 S46 tier-3, `func_80186E24`, 611 ins: 236-off "S11 regalloc-order" → MATCH, zero new pins) L10810 - **§175** — A CALLER-SAVED REGISTER PIN CAN BE A CORRECTNESS BUG, NOT JUST A SCHEDULING CHOICE (P31 wave H, 2026-08-15) L16780 - **§176j** — STOPPING A WAVE MID-FLIGHT COSTS THE IN-FLIGHT TAIL (and how much is recoverable) L17164 - **§177** — 🔴 THE EPILOGUE RETURN-DELAY SLOT IS DECIDED BY YOUR SAVED-REGISTER SET, NOT BY SCHEDULING L17227 - **§179-A** — 🔴 A LOOP-WALKED POINTER **PARAMETER** HANDS ITS ARGUMENT REGISTER TO THE GIV (9 byte-proofs) L17381 - **§179-F** — PINNING A LOOP-WALKED POINTER IS A TOTAL OFF-SWITCH FOR STRENGTH REDUCTION L17577 - **§179-G** — 🟡 A PIN CAN **CREATE** A COMBINE `LOG_LINK` AND DELETE AN `andi` (sixth RC-5 channel, n=1) L17600 - **§176** — SEVEN LEVERS FROM THE P31 OVERNIGHT WAVES (2026-08-15): statement order, false regalloc, and the pin that fights back L17678 - **§176-B** — "REGALLOC-PERM, 1-4 instructions off" ⇒ it is usually NOT register allocation L17708 - **§176-C** — 🔴 WALL REFUTATION: a hard-register pin CANNOT schedule around a call, because of a genuine gcc-2.7.2 bug L17754 - **§3-18** — of 20 reconciled while keeping the match. The two that did not are mechanism, not effort. L18007 - **§186** — CROSS-JUMPING RUNS **AFTER** SCHEDULING, SO NO C-LEVEL BARRIER CAN STEER IT L18137 - **§186c** — WHERE A VALUE IS LOADED DECIDES WHICH ALLOCATOR OWNS IT, AND THEREFORE ITS REGISTER L18166 - **§193-B** — A NARROW CAST OF A WIDE PARAMETER COSTS ONE EXTRA INSTRUCTION WHEN ITS STATEMENT SITS AFTER AN INTERVENING `jal` — AND THE DECIDER IS `combine.c:929`'s CROSS-CALL GUARD, NOT REGISTER ALLOCATION L18554 - **§193-D** — A COMPILER-GENERATED ARGUMENT COPY IS A `optimize_reg_copy_1` TRIGGER: when the pointer's LAST use in a block precedes a call that takes the pointer *as a register*, sched1 hoists the implicit `move $aN,$sN` to the block top and local-alloc re-bases the WHOLE block's memory operands onto `$aN`. The only C dial is a label (the §165-24 goto-join) between the block and the call. L18626 - **§194-B** — A `addu $rA,$rB,$zero` copy feeding ≥2 `sh` stores is a SECOND, 16-BIT-DECLARED local — the width, not the clamp or the join liveness, is what keeps the copy (and func_80183094's pin + `"memory"` clobber are both provably inert) L19061 - **§194-C** — A CALLER-SAVED loop counter proves its live range crosses ZERO calls — so it cannot share a pseudo with any value LIVE ACROSS a call (but it may freely share one with values that merely sit between calls) L19099 - **§194-D** — Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication + attribution correction of §165-17, NOT a new argument-register law L19162 - **§194-G** — Reading the integer half of a 16.16 stack aggregate: REGISTER-LIVENESS, not the C spelling, decides `sra` vs `lhu` — and the break is count-neutral L19313 - **§194-H** — §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a branch fences the store that reads v, and when TWO independent stores compete for the one delay slot the fence picks the winner at ZERO length drift (a WIDTH/sh!=sw swap, not a nop). Corrects §167-42's reconstruction (pseudo arm → hard-reg arm) and supplies its first re-runnable A/B. L19356 - **§194-N** — §193-D's C dial is misstated: the lever is a SURVIVING CODE_LABEL (a label with a real incoming edge), not "a label between the block and the call" — a bare label, or a `goto L; L:` pair whose target is the next active insn, is deleted by jump1 (jump.c:663-669 → delete_insn → jump.c:3458-3461, and jump.c:243 for the bare case) long before sched1/local-alloc, and costs exactly zero bytes L19673 - **§195-B** — A CALL_INSN does not start a basic block in gcc-2.7.2 — so a call-crossing temp can be a LOCAL-alloc quantity (the missing precondition under §48-A2 / §52 / regalloc.md K8) L19839 - **§195-F** — fold-const.c:4825 canonicalises a `?:` whose THEN arm is zero (or constant against a non-constant ELSE) by inverting the condition and swapping the arms — so a ternary's written arm order is byte-inert there, and `c ? 0 : X` is unspellable: it always compiles as `!c ? X : 0` L20055 - **§197-C** — Fix A1 (operand order) cannot move a commutative destination whose .greg conflict set already contains BOTH operand hard registers — split the accumulate so the destination IS the load's pseudo L20698 - **§199-A** — §189-A's asm→source inference is byte-FALSE: an interloper between a split constant's `lui`/`ori` is a SCHEDULE fact, not a source fact — and the separator is the BIRTHING BOOST, not a "priority floor" (§189-A's split-timing half survives; its "no statement order / no pin" absolute and the candidate's own forward-scheduler narrative both fall) L20769 - **§199-B** — A permutation sweep that holds ANY statement fixed is not a sweep: the statement an agent pins as "obviously load-bearing" is the one carrying the signal, and the partial sweep returns a FLAT residual that reads as proof of order-invariance L20818 - **§199-E** — §189-A BOUNDED AND CORRECTED — the discriminator is INSN_PRIORITY, not "is the interloper a constant": an insn between a `lui`/`ori` pair proves NOTHING about the source spelling unless it TIES the `ori` on priority, and on the pinned `-mcpu=3000` triple a dependent load never does L20971 - **§205** — THE CHAINED ASSIGNMENT IS ITS OWN SCHEDULING DIAL: `*b = *a = v;` moves an argument copy that no local, no pin and no statement reorder will move (P31 S56) L22108 - **§208** — TWO NAMED LOCALS FOR ONE RELOADED EXPRESSION BUY TWO ALLOCNOS — the naming granularity owns the REGISTER SPLIT, not just the load count (P31 S58) L22253 - **§211** — HOIST THE LOOP INIT ABOVE THE DOMINATING GUARD: it fills the guard's delay slot AND flips the counter/pointer register pair (P31 S58) L22425 - **§215** — PIN ECONOMY: the twin's pins are NOT part of the shape, and §17's "pin every call-crossing value" is over-broad (P31 S58) L22611 - **§220** — THE PARAMETER ITSELF IS A REGALLOC DIAL: use it directly, prefer `s32` to `void*`, and place the save-copy AFTER the first call (P31 S58) L22789 - **§221** — A CONSTANT SHARED BY TWO STORES DIES AT THE CALL WHOSE DELAY SLOT REFILLS ITS REGISTER **(single observation — not yet cross-confirmed)** (P31 S58) L22829 - **§30** — addendum (P31 S58) — the `/s` grant closes a SCHEDULING residual and a REGISTER residual with one edit L23267 - **§244** — `volatile` IS A COUNTING INSTRUMENT, A STORE-ORDER PIN, AND MUST SOMETIMES BE ASYMMETRIC (P31 S58b) L23761 - **§248** — SPLIT THE LOAD FROM THE ARITHMETIC: A FUSED `g + K` DENIES THE CALLEE-SAVED REGISTER ITS DIRECT HOME (P31 S58b) L23917 - **§253** — POSTFIX `++` vs `+= 1` PICKS A DIFFERENT SCRATCH REGISTER **(single observation — not yet cross-confirmed)** (P31 S58b) L24063 - **§254** — THE DEAD PARAMETER IS A REGISTER-PLACEMENT TOOL **(single observation — not yet cross-confirmed)** (P31 S58b) L24077 - **§215** — addendum (P31 S58b) — PIN ECONOMY, PART 2: NINE REFINEMENTS L24434 - **§223** — addendum (P31 S58b) — FIVE MORE CONFIRMATIONS, AND THE CONSTANT-IN-`$v0` CASE L24546 - **§263** — A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCHEDULE (P31 S59, byte-proven) L24937 - **ADD-8** — → §245 addendum — THE RE-TIE BARRIER PINS CALL-ARG SETUP TO SOURCE ORDER (construct 6) L25213 - **§268** — A `register __asm__` PIN ON A CALL-CLOBBERED REGISTER IS HONORED EXACTLY WHEN THE PINNED RANGE CROSSES NO CALL (P31 S59c; three A/B'd cards, unifying §257-2's two) L25295 - **ADD-6** — → §220-addendum — THE FOURTH FACE: WHEN THE NAMED COPY ITSELF BUYS THE EXTRA CALLEE-SAVED, PIN THE COPY TO `$16` L25437 - **ADD-7** — → §229 addendum — THE VALUE FACE: A LOOP-STORE CONSTANT SPELLED AS A LITERAL HAS NO SOURCE POSITION; NAME IT TO PIN THE PROLOGUE INIT ORDER L25453 - **ADDENDUM** — to §82 — struct copies must stay MEMBER-WISE, not block-moved, to match a spill-slot save L25753 - **ADDENDUM** — to §136d-1 (RC-12, the `$0`-add / opaque-copy family) — two symptoms beyond "compare reads the wrong register" L25791 - **ADDENDUM** — to §164-64 — AN EMPTY CLOBBER ON AN ARGUMENT REGISTER CAN BE THE DELIBERATE FIX, NOT JUST THE ACCIDENTAL BUG L26048 - **§275** — THE LEFTOVER-REGISTER READ L26422 - **§NNN** — A DO-WHILE'S GUARD AND LATCH MUST REPEAT THE SAME BOUND EXPRESSION, NOT SHARE ONE LOCAL: THE REPETITION IS WHAT BUYS THE CSE'D COPY INTO A SECOND REGISTER (P31 S60; `func_8017F2A4`, ov_SC03_096, byte-proven 25/25) L26933 - **ADDENDUM** — to §176-B2 — in a micro-function with no long/short lifetime asymmetry, BOTH contending pseudos need their own hard-register pin L27374 - **ADDENDUM** — to §220 — REFERENCING THE RAW PARAMETER (NO NAMED COPY, NOT EVEN A PIN) LETS THE CALLEE-SAVED SPILL LAND IN THE FIRST CALL'S OWN DELAY SLOT L27490 - **ADDENDUM** — to §215 — FIFTH SHAPE: reused mask constants across two call-free merge sites each get their own whole-function hard-register pin, and a shared sub-expression at the second site must be its own statement L27575 - **ADDENDUM** — to §194-B (func_8017EB34, ov_SC03_117 — wave dk; distinct from the §74 co-pinning finding on the SAME function below) L27983 - **ADDENDUM** — §NNN — sharpens §162a3/§60b: A `sll $v0,16 / sltiu $v0,1` ZERO-TEST OF A JUST-DECREMENTED HALFWORD IS A SIGNED-TEMP WIDTH TELL OUTSIDE ANY SWITCH/RANGE-TEST CONTEXT L28303 - **§285** — PRE-INITIALIZING A VARIABLE WITH A SHARED CONSTANT BEFORE A BRANCH MAKES BOTH ARMS OF THE FOLLOWING if/else DESTRUCTIVELY REUSE THE SAME DESTINATION REGISTER FOR THEIR BITWISE RESULT (byte-proven; `func_801811F0`, ov_SC03_102, independently rediscovered across waves di/dj) (P31 S60; waves #, byte-proven) L29088 - **§287** — A STACK-FRAME HOLE BELOW TWO ADDRESS-TAKEN AGGREGATE LOCALS IS A LEADING PAD MEMBER OF ONE COMBINED STRUCT, NOT SEPARATE LOCALS OR A REGISTER PIN (P31 S60/dj; `func_8017D104`, ov_SC06_027, byte-proven 47/47) (P31 S60; waves #, byte-proven) L29096 - **§288** — A REGISTER PIN DECLARED UNINITIALIZED AND ASSIGNED ONLY AT ITS LATE, SOLE USE STILL FORCES THE FIXED REGISTER'S SAVE/RESTORE, AND THE ASSIGNMENT'S SOURCE POSITION CONTROLS WHERE THE VALUE MATERIALIZES (P31; `func_80186530`, ov_SC02_017, byte-proven, match_one MATCH re-verified) (P31 S60; waves #, byte-proven) L29100 - **§290** — A SINGLE STRENGTH-REDUCED GIV CAN DRIVE STORES TO SEVERAL DISTINCT RELOCATABLE SYMBOLS, EACH KEEPING ITS OWN `%hi`/`%lo` ANCHOR (P31 S60; waves #, byte-proven) L29108 - **ADDENDUM** — to §172b-1 (counter-type dial) — the sll's SOURCE REGISTER is the placement discriminator L29278 - **ADDENDUM** — to §282 — the ADDRESS-GIV face: write the walked address INLINE so strength reduction births the induction register at loop.c's own insertion point L29323 - **§310** — A TWO-OPERAND `subu`'s DESTINATION TIES ONLY TO THE OPERAND THAT IS BLOCK-LOCAL: SHARE *ONE* SIDE ACROSS BLOCKS TO CHOOSE WHICH OPERAND'S REGISTER IT LANDS IN (P31 S64 t5j-t5m; byte-proven func_8017F578) L30096 - **§312** — A BARE RELATIONAL IN AN `if` GIVES THE `slt` NO TARGET: ASSIGN THE COMPARISON INTO THE ALREADY-ALLOCATED VARIABLE TO CHOOSE ITS DESTINATION REGISTER, AT ZERO INSTRUCTION COST (P31 S64 t5o-t5r; byte-proven func_80180DCC) L30169 - **Addendum** — §312 addendum — naming the compare's dest is the vehicle, the hard-reg pin on that name is (func_80184A68) L30471 - **Addendum** — subu/sh dest reuses a pinned operand's dying register (func_8017F498) L30475 - **Addendum** — Addendum to §312 — the compare's named destination must itself carry a hard register: nami (func_80184A68) L30491 - **Addendum** — REGALLOC-PERM: one memory-clobber barrier insufficient — stack two (func_80182D1C) L30533 - **REFUTED** — claim — Dead call-arg register address-fold: no C-level lever found (func_8017ECAC) L30551 - **Addendum** — REGALLOC-PERM: the store reads the narrow copy's register — split the one narrow local by (func_801838CC) L30636 - **Addendum** — REGALLOC-PERM store in a cross-jump shared tail defeats §137's barrier (func_8017EDE8) L30640 - **§319** — N TEXTUALLY DUPLICATED `return v;` TAILS PUT THE LAST `or` AND THE RETURN-REGISTER MOVE IN ONE BASIC BLOCK, SO combine FOLDS THEM AND YOU ARE EXACTLY −1; A `goto done;` JOIN WITH REAL INCOMING EDGES RESTORES THE SEPARATE `addu $v0,$a0,$zero` (P31 S66 round4; byte-proven func_801809F0) L30681 - **Addendum** — LENGTH-DRIFT −1 on a coalesced-away copy: a CALLER-SAVED SOURCE pin can resurrect it, boun (func_8017D72C) L30724 - **Addendum** — The dying-pinned-register reuse on a sign-extend chain: route every later read through a s (func_80186868) L30753 - **Addendum** — Register-pinned helper pointer local regresses closeness; use plain local (func_80013CFC) L30767 - **§325** — REGALLOC: A SHARED SMALL CONSTANT STORED TWICE IN THE PRE-LOOP BLOCK IS A *local-alloc* $s-OCCUPANT THAT STEALS THE ARGUMENT ALLOCNO'S REGISTER — PIN THE ARGUMENT-DERIVED LOCAL, NOT THE CONSTANT (P31 S67; byte-proven ov_SC03_119/func_80184F18, 153 ins) L30997 - **§329** — fold-const NARROWS `(int)s16_var & 0xFFF` ONTO THE *RAW HImode PSEUDO*, BREAKING THE SIGN-EXTEND/MASK REGISTER TIE — A ZERO-BYTE WIDENING TEMP RESTORES IT (P31 S67; byte-proven ov_SC01_084/func_80183244, 157 ins) L31044 - **§334** — A RELOAD SPILL SLOT IS ROUNDED TO `BIGGEST_ALIGNMENT` (8B), SO ONE SPILLED 4-BYTE PSEUDO CAN GROW THE FRAME BY 16 (P31 S67; byte-proven ov_SC05_008/func_8017DF68, 82 -> 53 residual) L31139 - **§342** — A `void *` PARAMETER CAST TO ITS REAL TYPE IN A LOCAL IS **NOT** BYTE-NEUTRAL WHEN A LATER PARAMETER ALSO NEEDS A CALLEE-SAVED REGISTER (P31 S67; byte-proven ov_SC07_002/func_80181394, NEW LAW) L31267 - **§344** — RAISE A BIV'S global_alloc PRIORITY WITH A ZERO-BYTE REFERENCE INSTEAD OF PINNING IT; PINNING THE COUNTER KILLS LSR ENTIRELY (P31 S67; byte-proven ov_SC03_121/func_80180E64, 222 ins) L31306 - **§347** — LOOP REGISTER ASSIGNMENT IS A **DECLARATION-ORDER + LIVE-RANGE** DIAL: FIVE COMPOSABLE LEVERS, 178 -> 0 (P31 S67; byte-proven ov_SC06_029/func_8017EF34, 243 ins) L31352 - **§365** — PIN **BOTH** MASKS OR NEITHER (P31 S68; ov_SC01_000/func_8017E594, 357 ins) L31758 - **§368** — ★★★ — THE **RELOAD-REMAT CONSTANT**: REACH A REGISTER NO PIN CAN REACH (P31 S68; ov_SC03_105/func_80187A30, 339 ins, fable escalation closed 8 → 0 in ONE edit) L31802 - **§373** — ★★★ — THE **DEAD-RESET CSE-BREAKER**, THE PIN THAT BREAKS A sched2 ANTI-DEP, AND WHY AN `asm` CAN NEVER RAISE PRIORITY (P31 S68; byte-proven ov_SC06_010/func_8017E764, 438 ins, fable escalation closed 8 → 0) L31987 - **§3-2.** — A REGISTER PIN THAT DELETES A sched2 ANTI-DEPENDENCE L32007 - **§374** — A `register … __asm__("$30")` RESERVATION IS **NOT HONOURED** BY `move_movables` (P31 S68; main/func_80015608, 86 ins) L32036 - **§375** — AN `$a0`-`$a3` PIN USED LATE RELOCATES AN **EARLIER** OUTGOING-CALL USE OF THAT REGISTER (P31 S68; main/func_8005F0C8, 88 ins) L32052 - **§3-C.** — REGISTER ALLOCATION FROM C, WITHOUT PINS L33134 - **§410** — ★★★ — COPY THEN ACCUMULATE ON THE COPY: resolving the birthing-boost vs register-allocation dilemma (P31 S71; byte-proven `ov_SC04_015/func_8017EB78`, 98 ins) L33363 - **§417** — ★★★ — A REGISTER PIN CAN BLOCK `jump.c`'s SELECT COLLAPSE, AND UNPINNING THEN EXPOSES A `cse` SKIP-BLOCKS MERGE (P31 S71; byte-proven `ov_SC03_013/func_8017E6F4`, 182 ins) L33615 - **§419** — ★★★ — WHEN A PIN IS IMPOSSIBLE, WIN THE local-alloc DENSITY CONTEST INSTEAD (P31 S71; byte-proven `ov_SC01_000/func_8017DD04`, 297 ins) L33668 - **§467** — GLOBAL-ALLOC TIES BREAK ON *DECLARATION* ORDER, AND A COPIED CLOBBER LIST IS A DEFECT L35366 - **§469** — THE `MEM_IN_STRUCT_P` ALIAS UNLOCK (and §463's spill law, independently confirmed) L35424 - **§471** — A LAUNDER'S REAL COST IS AN ALLOCNO, AND `$t0` IS RELOAD'S (`main:func_80032A74`, 408 → 12) L35475 - **§476** — 🔴 A HARD-REGISTER PIN DESTROYS TWO THINGS COMBINE AND SCHED1 NEED (`func_800226C0`, 670 ins → MATCH) L35622 - **§463** — 🔴 SPILL SLOTS ARE 8 BYTES, AND THE §41b "LOAD ABOVE THE PROLOGUE" WALL IS REFUTED L35687 - **§478** — 🔴 — A VERBATIM DRAFT IS THE STRONGEST *FALSE* SIGNAL YOUR SCOPING TOOL CAN EMIT L35797 - **§499** — ★★★ — A NEVER-ONBOARDED PAYLOAD PINS ITS OWN BASE STATICALLY, AND THE FIRST BUILD CANNOT (P32 T2a/T2b: the parked five onboarded, 213 → 218 binaries, 0 UNCLAIMED) L36758 - **§501** — ★★★ — A LEVER THAT MEASURES WORSE MAY BE A CASCADE: READ THE `.loop` DUMP FOR THE DESIRABILITY FLIP BEFORE DISCARDING IT (P32 T4b, `main:func_800391D4`, a pinned wall banked by a Fable agent) L37157 - **§456** — The lever-removal move catalog: what the S104 agents found under ~205 pins, barriers and launders (Phase 36 T7) L37765 ### CSE / redundancy / rematerialization (50) - **§46** — The `func_80178D40` crack (890 ins ×134, the heaviest core in the game): four LOOP-STRUCTURE levers cheap-Opus found by reading loop.c/jump.c/cse.c (Phase 26 session 8, 2026-07-13) L3351 - **§83d** — CSE's quantity budget is WHOLE-FUNCTION, so a local rewrite cannot fix a local symptom L6490 - **§153** — THE ADDRESS-REMATERIALISATION LAUNDER: a third zero-emission asm lever (P30 S43, `func_8018D98C`, 710 ins) L10504 - **§176-D** — CSE-class levers used in reverse (two sharpenings of §153 and cse_expr §2) L17780 - **§193-E** — A varying-address (pointer) load is re-emitted once per CSE-LIVE INTERVAL, and naming it in a C local is the only C-level lever over that count — no store SPELLING has any reach L18697 - **§193-F** — §148-A2 — The `threshold -= 3` STAIRCASE: `move_movables` admits a COUNT of invariants, not a boolean — two identical merged constants split hoisted/not-hoisted by LIST ORDER, and `insn_count` picks the rank cutoff L18727 - **§193-H** — A pointer-derived base load `*(s32*)(p+K)` is uncacheable across ANY memory write or call (cse's third kill disjunct) — so the target emits one `lw` per store-separated RUN of spellings, and a run of stores sharing one `lw` is a C local you must declare L18852 - **§194-A** — A zero-byte scheduling fence goes AFTER the defining statement to make that computation emit FIRST in its block — and the barrier predicate is `volatile`-or-colon-less, not the `"memory"` clobber (COMPLEMENTS §165-40; does not refute it) L18995 - **§194-J** — Back-to-back identical stores: flow.c's `last_mem_set` deletes the first, and only `volatile` saves it L19437 - **§194-K** — Blind sched1's alias oracle with a second SET of a pointer pseudo — the first zero-byte, non-volatile, dependence-CREATING lever (corrects §167-05's "volatile is the only door"; fourth consumer of reg_n_sets) L19489 - **§195-H** — §165-27g — ON A FIXED-SYMBOL GLOBAL, `extern T D[];` + `D[0]` IS THE CSE RELOAD DIAL, AT ZERO ADDRESSING COST, AND THE ELEMENT COUNT IS INERT (§165-27's `T v[2]` CAVEAT IS A LOCAL-FRAME FACT AND DOES NOT TRANSFER) L20155 - **§195-I** — §145(b) AMENDED — the pointer-bump `addiu` is saved by cse, not combine: ANY second SET of a pseudo in the address's equivalence chain (in-place `p += K`, §145(b)'s `p = r;` copy, or an asm re-tie) kills the fold; the pass is cse and the gate is `invalidate`'s `reg_tick++` L20202 - **§195-L** — The cse store-re-seed does not cross a JOIN LABEL: per-arm stores + a join read keep the reload that one join store deletes (bounds §193-E BOUND 1/BOUND 3 with §48-B's EBB boundary) L20363 - **§197-A** — A NARROW SIGNED MEMORY READ FEEDING A CONSTANT `>>` LOSES ITS `lh`, AND THE CURE IS AN ASM RE-TIE (attribution CONTESTED: cse vs combine) L20626 - **§197-B** — A REPEATED COMPARE OF ONE VALUE AGAINST ONE CONSTANT IS DELETED BY cse's `qty_comparison_code` CHANNEL (the non-EQ complement of §165-03) — and a front-end-opaque mask on EITHER compare is a pure-C dial that keeps the target's second branch L20670 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§201-E** — §194-J-2 — The `last_mem_set` deletion window is measured in SOURCE/expand order, not in the printed stream; a foreign store moved between the pair in C source is a real lever, and `volatile` is not always the better one L21419 - **§204-D** — A LOOP-INVARIANT LOAD IS ADMITTED AS A MOVABLE ONLY IF ITS ADDRESS CANNOT TRAP: `local.field` PASSES, THE SAME READ THROUGH A POINTER LOCAL IS REFUSED L21814 - **§243** — SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P31 S58b) L23731 - **§244** — `volatile` IS A COUNTING INSTRUMENT, A STORE-ORDER PIN, AND MUST SOMETIMES BE ASYMMETRIC (P31 S58b) L23761 - **ADD-1** — → §257-8 addendum — THE INTERPOSED ASM'S `__volatile__` IS A PER-SHAPE DIAL, AND THE "NO-OUTPUT ASM IS IMPLICITLY VOLATILE" LORE IS BYTE-FALSE IN gcc-2.7.2 L25364 - **ADDENDUM** — to §244 — THE COMPILER BARRIER, NOT `volatile`, BLOCKS A STORE→LOAD HOIST THROUGH A POINTER-OFFSET FIELD L26079 - **ADDENDUM** — to §211 — AN IN-LOOP ACCUMULATOR WANTS A CLOSED-FORM EXPRESSION WHEN THE TARGET REMATERIALISES ITS CONSTANT AFTER EVERY CALL L26387 - **§276** — MIXED ADDRESS-EXPRESSION SPELLING FOR ADJACENT RELOCATABLE SYMBOLS IS A CSE-UNIFICATION DIAL, NOT JUST A BYTE-ENCODING CHOICE (P31 S60; `func_80180FE8`, ov_SC06_006, byte-proven) L26462 - **§NNN** — A DO-WHILE'S GUARD AND LATCH MUST REPEAT THE SAME BOUND EXPRESSION, NOT SHARE ONE LOCAL: THE REPETITION IS WHAT BUYS THE CSE'D COPY INTO A SECOND REGISTER (P31 S60; `func_8017F2A4`, ov_SC03_096, byte-proven 25/25) L26933 - **ADDENDUM** — to §22 (`volatile`-qualified-global reload lever, cookbook ~L1922) — for a NON-constant, same-address double RMW, a plain memory clobber beats `volatile`, and `volatile` actively breaks a delay-slot fill L27236 - **§302** — A RED BINARY IS A DRIFTED SPEC, NOT A MYSTERY: THE THREE CARVE-STATE INVARIANTS AND HOW TO DERIVE EACH FROM THE BYTES (P31 S62 T2; five reds healed in one session, 5/5, +39 held banks) L29729 - **§308** — A NAIVE FINAL `goto` DELETES THE TARGET'S `j`+`nop`: SPELL THAT EDGE AS A TAUTOLOGICAL `if` SO cse FOLDS IT TO AN UNCONDITIONAL JUMP *AFTER* jump1's WINDOW HAS CLOSED (P31 S63 t5e-t5i; byte-proven func_80180808) L29993 - **§317** — A NARROW STRUCT-FIELD STORE TURNS A *LIVE SImode VARIABLE* INTO A `li`: convert.c's TRUNCATION-DISTRIBUTION MINTS A HImode COPY, AND cse REWRITES THAT COPY'S SOURCE TO THE EQUIVALENT CONSTANT (P31 S66; byte-proven func_8017EF94) L30354 - **Addendum** — Addendum to §167-12 — the single-operand volatile keepalive, re-confirmed at a gated MATCH (func_8017F498) L30495 - **REFUTED** — claim — REFUTED — the `j`-slot store is an ASPSX macro-expansion hop, not a cse split; maspsx hard (func_8005DBD8) L30599 - **§NNN** — REFUTED: A `sw %lo(SYM)($at)` IN A JUMP'S DELAY SLOT WITH ITS `lui $at` STRANDED ABOVE THE JUMP IS **ASSEMBLER** MACRO-EXPANSION, NOT A cse SPLIT — OUR PIPELINE CANNOT EMIT IT FROM ANY C, AND THE +1 IS A `maspsx` GAP (P31 S66 round2, `func_8005DBD8`, byte-probed) L30601 - **Addendum** — Volatile zero-byte asm slider scrambles the prologue schedule (func_801814B0) L30757 - **§326** — DEFEATING ADDRESS-CSE: SPELL TWO READS OF THE SAME HALFWORD DIFFERENTLY AND GCC CANNOT SHARE THE ADDRESS (P31 S67; byte-proven ov_SC01_077/func_8017FAAC, 154 ins) L31010 - **§327** — A RANGE TEST MUST BE HImode: WITH `s32` + A `(u16)` CAST GCC *PROVES* THE MASK REDUNDANT AND DROPS THE `andi` (P31 S67; byte-proven ov_SC02_039/func_8017EC34, 154 ins) L31019 - **§328** — THE VOLATILE ALIAS MUST BE AN *OBJECT*, NOT A CAST: `*(volatile s32*)&sym` UNFOLDS %lo INTO A SEPARATE `addiu` (P31 S67; byte-proven ov_SC07_007/func_80181B8C, 156 ins, NEW LAW) L31034 - **§345** — A VOLATILE **STORE** EVICTS THE MEM FROM cse AND KEEPS `sh`; A VOLATILE **LOAD** BLOCKS combine AND DEGRADES `lh` INTO `lhu+sll+sra` (P31 S67; byte-proven ov_SC01_084/func_80181A7C) L31322 - **§349** — RE-ASSIGN A BASE POINTER AT THE END OF THE LOOP BODY TO MAKE `n_times_set > 1` — THAT KILLS BOTH loop.c's INVARIANT HOIST **AND** THE ADDRESS GIV, REPRODUCING A REMATERIALISED `addiu $aN,$sp,K` (P31 S67; byte-proven ov_SC06_029/func_801804C8, 255 ins) L31424 - **§351** — `/s` (MEM_IN_STRUCT_P) IS A DIAL YOU CHOOSE PER ACCESS: A COMPONENT_REF GRANTS IT AND LETS cse KEEP AN INDEX ACROSS THE STORE; A PLAIN CAST DENIES IT (P31 S67; ov_SC01_000/func_8017DD04, 186 -> 5) L31466 - **§352** — ⚠ TWO IDENTICAL `__asm__ __volatile__("")` BARRIERS **MERGE WITH EACH OTHER** — SPELL THE SECOND ONE DIFFERENTLY (P31 S67; byte-proven resident/func_800D128C, measured closeness 105 when they merged) L31496 - **§368** — ★★★ — THE **RELOAD-REMAT CONSTANT**: REACH A REGISTER NO PIN CAN REACH (P31 S68; ov_SC03_105/func_80187A30, 339 ins, fable escalation closed 8 → 0 in ONE edit) L31802 - **§373** — ★★★ — THE **DEAD-RESET CSE-BREAKER**, THE PIN THAT BREAKS A sched2 ANTI-DEP, AND WHY AN `asm` CAN NEVER RAISE PRIORITY (P31 S68; byte-proven ov_SC06_010/func_8017E764, 438 ins, fable escalation closed 8 → 0) L31987 - **§3-1.** — DEAD-RESET CSE-BREAKER — the zero-footprint replacement for a §195-I asm re-tie L31989 - **§385** — ★★★ — THE **SCHED2 PRIORITY-DONOR ASM**: closing the "hoisted-invariant vs IV-init preheader swap" class (P31 S69; byte-proven main/func_80038A58, 347 ins, fable escalation 2 → 0) L32319 - **§417** — ★★★ — A REGISTER PIN CAN BLOCK `jump.c`'s SELECT COLLAPSE, AND UNPINNING THEN EXPOSES A `cse` SKIP-BLOCKS MERGE (P31 S71; byte-proven `ov_SC03_013/func_8017E6F4`, 182 ins) L33615 - **§432** — ★★★ — DEFEAT cse's MERGE OF TWO IDENTICAL MASKS BY SPELLING ONE AS A SHIFT PAIR (P31 S72/S73; `main/func_8002DC68`, MATCH 198/198) L34227 - **§461** — LAUNDERING AN INVARIANT CAN BE THE DEFECT, AND "RESIDUAL A" DOES NOT GENERALIZE PAST ONE BINARY OP L35264 - **§464** — FOUR VOLATILE/BARRIER LEVERS FROM `main:func_8005DE78` (141 ins → MATCH) L35338 - **§470** — FOUR CSE/SCHED LEVERS FROM `main:func_800301C8` (170 ins, 133 → 18) L35447 - **§475** — THE `"memory"` FENCE AS A *CSE INVALIDATOR*, AND `(b*3)<<3` INSTEAD OF `b*24` L35591 ### loops & induction variables (50) - **§3-T1** — Loop pointer: top-of-body for `addu` induction, not constant-folded `addiu` L71 - **§34** — The `func_80138ED0` giant crack: gcc-2.7.2's **3-qty sort bug** + the **zero-byte asm allocation toolkit** + the **giv-init fence** (Phase 24 T5; Opus→close=21, Fable5→MATCH ×134) L2477 - **§35** — The region-a sibling-giant harvest: difficulty ≠ $s-reg count (it's global-array hoisting) + the banking recipe + new loop idioms (Phase 24 T7, 5 parallel Opus-Max agents) L2494 - **§46** — The `func_80178D40` crack (890 ins ×134, the heaviest core in the game): four LOOP-STRUCTURE levers cheap-Opus found by reading loop.c/jump.c/cse.c (Phase 26 session 8, 2026-07-13) L3351 - **§66d** — The permuter⇄reader loop: alternate a random search with a byte-verified idiom, and let `residual_class` decide whose turn it is (Phase 29 SESSION-17) L5283 - **§66d-1** — What transfers between giants is the LOOP, not the PIN L5318 - **§70** — The giv-init base register: walk the PARAMETER, not a copy of it (Phase 29 SESSION-18, `func_801777BC`) L5657 - **§145** — Three loop/combine levers from the S40 wave-2 drafters (16/16 match_one) L9965 - **§148** — The loop.c hoisting THRESHOLD is arithmetic you can compute, and the `?:` clamp that folds to MIN_EXPR (P30 S42, `func_8017C6F4`, 947 ins) L10183 - **§171a** — THE MECHANICAL A-PROP DRAFT (P30 S50): 256 members banked with no agent in the loop L16478 - **§179-A** — 🔴 A LOOP-WALKED POINTER **PARAMETER** HANDS ITS ARGUMENT REGISTER TO THE GIV (9 byte-proofs) L17381 - **§179-E** — A `>2*MAX_MOVE_BYTES` BLOCK COPY IS A **STRUCT ASSIGNMENT**, NOT A HAND LOOP L17551 - **§179-F** — PINNING A LOOP-WALKED POINTER IS A TOTAL OFF-SWITCH FOR STRENGTH REDUCTION L17577 - **§194-C** — A CALLER-SAVED loop counter proves its live range crosses ZERO calls — so it cannot share a pseudo with any value LIVE ACROSS a call (but it may freely share one with values that merely sit between calls) L19099 - **§204-B** — A LOOP COUNT THAT ARRIVES ON THE STACK IS DECREMENTED IN PLACE: a fresh counter local can cost a real `move` AND permute the whole callee-saved file L21692 - **§204-D** — A LOOP-INVARIANT LOAD IS ADMITTED AS A MOVABLE ONLY IF ITS ADDRESS CANNOT TRAP: `local.field` PASSES, THE SAME READ THROUGH A POINTER LOCAL IS REFUSED L21814 - **§211** — HOIST THE LOOP INIT ABOVE THE DOMINATING GUARD: it fills the guard's delay slot AND flips the counter/pointer register pair (P31 S58) L22425 - **§246** — THREE-LIVE-VALUE SCAN LOOPS WANT ADDRESS-FROM-INDEX; AND TWO SYMBOLS CAN SHARE ONE giv (P31 S58b) L23843 - **ADD-7** — → §229 addendum — THE VALUE FACE: A LOOP-STORE CONSTANT SPELLED AS A LITERAL HAS NO SOURCE POSITION; NAME IT TO PIN THE PROLOGUE INIT ORDER L25453 - **ADD-8** — → §30a addendum — THE THIRD COLUMN: `*p++` IS ALSO SERIALIZED BUT BURNS `addiu`; ONLY CAST-PLUS GIVES SERIALIZATION *AND* DISPLACEMENT FOLDING L25466 - **ADDENDUM** — to §172b-1 — TWO PLACEMENT DIALS THE PROMOTION LAW DOESN'T NAME: NARROW THE COUNTER'S DECLARED TYPE, AND MOVE THE CAST INTO THE LOOP BODY L26110 - **ADDENDUM** — to §211 — AN IN-LOOP ACCUMULATOR WANTS A CLOSED-FORM EXPRESSION WHEN THE TARGET REMATERIALISES ITS CONSTANT AFTER EVERY CALL L26387 - **§279** — A `do/while (p < end)` LOOP UNDER AN ENTRY GUARD: the guard decides the compare, not the loop (P31 S60; wave cf, func_8017F2A4, ov_SC03_096, byte-proven) L26932 - **§NNN** — A DO-WHILE'S GUARD AND LATCH MUST REPEAT THE SAME BOUND EXPRESSION, NOT SHARE ONE LOCAL: THE REPETITION IS WHAT BUYS THE CSE'D COPY INTO A SECOND REGISTER (P31 S60; `func_8017F2A4`, ov_SC03_096, byte-proven 25/25) L26933 - **§NNN** — A LOOP CURSOR'S C TYPE (POINTER vs PLAIN INTEGER) SELECTS `sltu` vs `slt` FOR ITS BOUND TEST, INDEPENDENT OF THE VALUES INVOLVED (P31 S60; `func_800CAE74`, md_MAIN_031, byte-proven; cross-confirmed same wave by `func_8017F2A4`, ov_SC03_096) L26967 - **§282** — gcc's OWN LOOP REVERSAL PUTS THE COUNTER INIT AFTER THE HOISTED MOVABLES — a position no hand-written down-count can reach (P31 S60; wave cf, func_80180DD8, ov_SC04_005, byte-proven) L27046 - **§NNN** — WRITE THE UP-COUNT LOOP: gcc's OWN REVERSAL PRODUCES A COUNTER-INIT INSTRUCTION THAT LANDS AFTER HOISTED MOVABLES, WHICH A HAND-WRITTEN DOWN-COUNT LOOP CANNOT REPRODUCE (P31 S60; `func_80180DD8`, ov_SC04_005, byte-proven) L27047 - **§290** — A SINGLE STRENGTH-REDUCED GIV CAN DRIVE STORES TO SEVERAL DISTINCT RELOCATABLE SYMBOLS, EACH KEEPING ITS OWN `%hi`/`%lo` ANCHOR (P31 S60; waves #, byte-proven) L29108 - **ADDENDUM** — to §282 — the ADDRESS-GIV face: write the walked address INLINE so strength reduction births the induction register at loop.c's own insertion point L29323 - **ADDENDUM** — to the L1800 anchor-steer bullet — a DERIVED-POINTER local silently flips the merged giv's anchor END L29367 - **§297** — ONE GIV SERVING MIXED-WIDTH LOADS *AND* A STORE NEEDS STRUCT-MEMBER SPELLING; CAST/INDEX SPELLINGS RE-SPLIT THE PSEUDO (P31 S61; wave m0a, `func_800347C8`, main, byte-proven 31/31) L29509 - **§312** — A BARE RELATIONAL IN AN `if` GIVES THE `slt` NO TARGET: ASSIGN THE COMPARISON INTO THE ALREADY-ALLOCATED VARIABLE TO CHOOSE ITS DESTINATION REGISTER, AT ZERO INSTRUCTION COST (P31 S64 t5o-t5r; byte-proven func_80180DCC) L30169 - **Addendum** — Loop sentinel wrongly hoisted out of a call-containing loop (func_801835B0) L30503 - **§325** — REGALLOC: A SHARED SMALL CONSTANT STORED TWICE IN THE PRE-LOOP BLOCK IS A *local-alloc* $s-OCCUPANT THAT STEALS THE ARGUMENT ALLOCNO'S REGISTER — PIN THE ARGUMENT-DERIVED LOCAL, NOT THE CONSTANT (P31 S67; byte-proven ov_SC03_119/func_80184F18, 153 ins) L30997 - **§344** — RAISE A BIV'S global_alloc PRIORITY WITH A ZERO-BYTE REFERENCE INSTEAD OF PINNING IT; PINNING THE COUNTER KILLS LSR ENTIRELY (P31 S67; byte-proven ov_SC03_121/func_80180E64, 222 ins) L31306 - **§346** — `c ? X : -X` TAKES expand_expr's COND_EXPR **SINGLETON** PATH (copy, then negate IN PLACE) — AN if/else STATEMENT GIVES THE TWO-ARM FORM (P31 S67; byte-proven ov_SC03_102/func_80180C38, closed the last instruction) L31332 - **§347** — LOOP REGISTER ASSIGNMENT IS A **DECLARATION-ORDER + LIVE-RANGE** DIAL: FIVE COMPOSABLE LEVERS, 178 -> 0 (P31 S67; byte-proven ov_SC06_029/func_8017EF34, 243 ins) L31352 - **§348** — THE BASE SPELLING PICKS THE ADDRESSING MODE: A SYMBOL GIVES THE 3-INSN `lui/%lo` FORM, A POINTER VARIABLE GIVES THE 2-INSN `addu/lw` FORM (P31 S67; byte-proven ov_SC07_007/func_80182184, 264 -> 30 on this row alone) L31407 - **§349** — RE-ASSIGN A BASE POINTER AT THE END OF THE LOOP BODY TO MAKE `n_times_set > 1` — THAT KILLS BOTH loop.c's INVARIANT HOIST **AND** THE ADDRESS GIV, REPRODUCING A REMATERIALISED `addiu $aN,$sp,K` (P31 S67; byte-proven ov_SC06_029/func_801804C8, 255 ins) L31424 - **§354** — THE giv **WORTH-WHILE TEST** IS A DIAL: RE-ASSOCIATE THE ADDEND INTO THE INDEX TERM AND `strength_reduce` DECLINES (P31 S68; byte-proven ov_SC03_105/func_801824CC, 320 ins, 201 → 5 → 0) L31552 - **§357** — ONE STRUCT POINTER, NOT TWO: A SECOND SOURCE VARIABLE BUILDS A THIRD IV (P31 S68; byte-proven ov_SC06_029/func_80181DF8, 335 ins, 330 → 13) L31599 - **§361** — ★ — A LOOP-TAIL BYTE SIGNATURE THAT NAMES ITS SOURCE SHAPE — AND THE LAW THAT A "SCHEDULING TIE" MAY BE YOUR OWN EARLIER LEVER (P31 S68; byte-proven main/func_800241C0, fable escalation, 19 → 0 in 3 iterations) L31647 - **§366** — ★★ — `group_case_nodes` MERGES **STACKED CONSECUTIVE** CASE LABELS: GIVE EVERY CASE ITS OWN BODY (P31 S68; ov_SC01_001/func_8017EC28, **first-try MATCH 360/360**, 96/96 relocs audited) L31767 - **§385** — ★★★ — THE **SCHED2 PRIORITY-DONOR ASM**: closing the "hoisted-invariant vs IV-init preheader swap" class (P31 S69; byte-proven main/func_80038A58, 347 ins, fable escalation 2 → 0) L32319 - **§386** — ★★★ — A BYTE LOAD ON THE **BIV** BASE WAS BORN IN THE COMBINE PASS: SPELL IT AS A SHIFT-MASK, NEVER A DEREF (P31 S69; byte-proven main/func_80020598, 292 ins, escalation 1 → 0) L32345 - **§393** — ★★ — THE **BIRTHING BOOST**: A SINGLE-SET LOCAL IS SCHEDULED LAST; GIVE IT A SECOND SET (P31 S69; byte-proven ov_SC02_017/func_8017FCFC) L32553 - **§418** — ★★★ — TWO LOOP-STRUCTURE LEVERS: MAKE THE SECOND INDEX A GIV, AND KEEP A TABLE ADDRESS UNFOLDED (P31 S71; byte-proven `ov_SC04_016/func_8017DF8C`, 184 ins, 32 → 0 in seven compiles) L33642 - **§430** — ★★★ — A GOTO INTO A LOOP IS FINE; HAND-HOIST THE CONSTANTS IT COSTS YOU (P31 S73 — **this section previously said the OPPOSITE and was wrong; the refutation is kept below**) L34083 - **§472** — 🔴 §148-A's HOIST THRESHOLD IS 29, NOT 58, WHEN THE LOOP CONTAINS A CALL L35499 - **§501** — ★★★ — A LEVER THAT MEASURES WORSE MAY BE A CASCADE: READ THE `.loop` DUMP FOR THE DESIRABILITY FLIP BEFORE DISCARDING IT (P32 T4b, `main:func_800391D4`, a pinned wall banked by a Fable agent) L37157 ### structs, block moves & memcpy (93) - **§3-T2** — Source statement order drives instruction scheduling L78 - **§5** — Known hard-residual classes (instruction-identical, one byte-exact blocker) L199 - **§13** — Add a location overlay — the canonical runbook (Phase 13; the Phase-15 fleet recipe) L1075 - **§15** — Struct-heavy shared-core pipeline (Phase 16) — empirical determinations (S0) L1336 - **§16** — Guided hand-matching the struct-heavy core (Phase 17 — beats the §15 brute-force) L1358 - **§3-The** — crack: the residual was gcc-2.7.2 COPY-COALESCING, not the instruction scheduler L2167 - **§28** — Banking a "close=0 gate-rejected" giant: the canonical-extern recovery (Phase 22 T1, byte-proven on func_8015126C ×134) L2348 - **§28b** — The struct-walled close=0 giant (§28 case #3) IS bankable: the engine_types.h type-lift (Phase 22 T2, byte-proven on func_80156B74 ×134) L2380 - **§30** — Fable5Max cracks §20/§10 "unsteerable" from the gcc SOURCE: store-vs-load is a `/s` aliasing flag, the def-side wall has a macro escape, + the birthing-boost (Phase 23, byte-proven on giant `func_8014EE14` 248 ins ×134) L2405 - **§32** — The region-a CAMERA-GIANT idiom set: struct-base hoisting + 4 sibling levers (Phase 24 T7, Fable5-cracked on `func_80129CF8` 191 ins, match_one MATCH; transferable to the 6 sibling giants) L2445 - **§38** — The WHALE `func_80144B9C` (770 ins): the -O0 struct-assign memcpy idiom + the -O0 reach-134 ×134 rollout (Phase 24 T7 §G, cheap Opus — no Fable5, no calls.c) L2550 - **§40** — Structural families: the MECHANICAL symbol-remap (crack one exemplar → remap the rest, ~0 tokens) (Phase 25 T3/T7, 2026-07-08) L2576 - **§41** — The DEF-SIDE canonical-sig wall: mechanically banking a drafted giant past `conflicting types` (Phase 25 T5b batch-2, 2026-07-09; `tools/canon_sig_reconcile.py`, byte-proven on `func_8013B274`) L2733 - **§42a** — addendum — wave 2 (residuals + 29-100 band): iso-MATCH ≠ real-TU bank, the memcpy→struct-assign fix, +5 levers (2026-07-10b) L2991 - **§44** — The Phase-25 cheap-Opus giant batch: 5 structural levers + the §43 extension (2026-07-11, 6 crackers over the frontier giants) L3280 - **§46** — The `func_80178D40` crack (890 ins ×134, the heaviest core in the game): four LOOP-STRUCTURE levers cheap-Opus found by reading loop.c/jump.c/cse.c (Phase 26 session 8, 2026-07-13) L3351 - **§3-Why** — 0/8 was structural, and predictable from two words L4083 - **§54** — `--fix-def-sig`: the member's CANONICAL DECLARATION is a build step too (tiny-IMM mega-pools, +4,801, Phase 29 T6, 2026-07-16) L4201 - **§56b** — PROPAGATING an h_seq giant: the exemplar's externs MUST be fleet-canonical, not the draft's types L4270 - **§65a** — The blast-radius taxonomy (makes §61's law structural instead of remembered) L5068 - **§66d-5** — `residual_class`'s "structural ⇒ permuter CPU is waste" is WRONG for schedule permutations (measured, Phase 29 SESSION-18) L5529 - **§83** — The parameterised-repeat law, the spill-area trap, and why a per-case edit cannot move a per-case symptom (Phase 29 SESSION-20, `func_80183814` 5,122 ins, cold-ish start → 36 structural / 99.3%) L6450 - **§84** — The DERIVED-OFFSET remap bug: a hand-computed literal that encodes the DISTANCE between two per-overlay symbols, and why `match_one` is structurally blind to it (Phase 29 SESSION-20, `func_8013D53C`) L6510 - **§3-The** — construct L6515 - **§91** — A structure-TRANSFER is only valid where the structure corresponds: the `--like` role trap (Phase 29 SESSION-21, `func_8012AAAC` ×137) L6880 - **§3-The** — three-hypothesis trail, because two of them were wrong and the wrongness is instructive L6909 - **§129** — Post-carve, `rtu_match`/`match_one` COUNT THE JUMP TABLE AS INSTRUCTIONS; and a carve must never be committed without its owner (P30 S28, `func_8013BD74`) L8475 - **§129a** — the target instruction count is INFLATED after a carve L8479 - **§3-The** — DEFINITION-side alias is the only escape when the fleet canon disagrees on a promoting param L9511 - **§3-Two** — errors of mine, both instructive L10047 - **§160** — THE REACH-15 WAVE HARVEST: an align-1 block move, and the instrument that called it a wall (P30 S47) L10931 - **§16Xy** — SHARPENS *(sharpens §136d-3, §37 /s-DEP LATTICE, §135-2, §136-13, §162q)* L12338 - **§3-The** — `/s` drop clause is in ALL THREE dependence predicates, so the FIXED-ADDRESS STORE is what floats (P30 S48, `func_8017CA18`, ov_MAIN_012) L12340 - **§16Z** — SHARPENS *(sharpens §37 "the /s-DEP LATTICE", §136-13, §136-14, §16Xy, §162q, `gcc-2.7.2-map/sched.md` §64)* L14398 - **§3-The** — ADDRESS-CLASS TABLE: which load/store pairs even REACH the `/s` clause (P30 S48 wave 4, `func_80185B44`, ov_SC03_014) L14400 - **§176g** — SIZE A WAVE BY INSTRUCTIONS, NOT BY CARDS (P31 S52 — the adopted doctrine) L16998 - **§3-F.** — `MEM_IN_STRUCT_P` ASYMMETRY IN `true_dependence` (func_80037144, 124 ins) L17349 - **§179-E** — A `>2*MAX_MOVE_BYTES` BLOCK COPY IS A **STRUCT ASSIGNMENT**, NOT A HAND LOOP L17551 - **§176-B** — "REGALLOC-PERM, 1-4 instructions off" ⇒ it is usually NOT register allocation L17708 - **§189** — FIVE COMPILER LAWS MINED FROM THE WAVE R/S JOURNALS (P31 S53), each source-cited and re-derived by a second agent L18271 - **§193-A** — The wave card ships only OPEN-set pointers and discards the atlas's BANKED one — `exemplar`/`sibs` are stubs 100% by construction (tools/atlas.py:96/657), while `seed.ref` (matched pool, atlas.py:505-536) is dropped at build_wave_atlas.py:143 L18487 - **§193-B** — A NARROW CAST OF A WIDE PARAMETER COSTS ONE EXTRA INSTRUCTION WHEN ITS STATEMENT SITS AFTER AN INTERVENING `jal` — AND THE DECIDER IS `combine.c:929`'s CROSS-CALL GUARD, NOT REGISTER ALLOCATION L18554 - **§193-I** — A DECLARED AGGREGATE LOCAL HAS AN 8-BYTE FRAME STRIDE — CEIL(size,8), NOT size. N ARRAYS THEREFORE COST 8·k MORE THAN ONE HAND-LAID STRUCT (k = how many of them have size mod 8 ≠ 0), AND THE TELL IS IDENTICAL INSTRUCTION COUNT WITH EVERY $sp DISPLACEMENT SHIFTED BY THE SAME CONSTANT. L18892 - **§194-E** — The wave card's `exemplar` is the TARGET ITSELF on 42/73 wave-U and 36/71 wave-T cards — a construction consequence of `atlas.py:657` (exemplar = max-nins OPEN member) meeting `build_wave_atlas.py:166` (one card per gid); and the shipped §193-A `seed_ref` is same-binary on 0/51, so no card field can ever name a destination-TU sibling L19203 - **§194-H** — §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a branch fences the store that reads v, and when TWO independent stores compete for the one delay slot the fence picks the winner at ZERO length drift (a WIDTH/sh!=sw swap, not a nop). Corrects §167-42's reconstruction (pseudo arm → hard-reg arm) and supplies its first re-runnable A/B. L19356 - **§195-D** — §195 — `masked_diff.mask_for` returns 0 for EVERY `j`/`jal` word, so an internal `j` destination is invisible to match_one, the permuter scorer AND every similarity tier: a control-flow semantic error (which calls execute) surfaces as a 1-instruction residual mislabelled `DELAY-SLOT / profile=schedule`, or as an outright false MATCH L19929 - **§195-F** — fold-const.c:4825 canonicalises a `?:` whose THEN arm is zero (or constant against a non-constant ELSE) by inverting the condition and swapping the arms — so a ternary's written arm order is byte-inert there, and `c ? 0 : X` is unspellable: it always compiles as `!c ? X : 0` L20055 - **§195-G** — §NEW — TWO ARMS CALLING THE SAME CALLEE MERGE INTO ONE `jal` UNLESS EACH ARM'S OWN CODE CONSUMES THE RESULT: the C dial that keeps two call sites apart is WHERE the consumer test lives, and the branch SENSE you spell it with is byte-inert (jump.c:1737 canonicalises it) L20107 - **§195-K** — At -O2 the §18 array-of-struct lever is a FRAME lever, not a length lever — but only when the N reads carry DISTINCT index expressions; with a SHARED index §18's +2-instruction residual is still alive at -O2 (the submitted "memory-loaded narrow index" precondition and unconditional length-neutrality are both falsified) L20311 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§199-F** — §164-36b — THE TARGET-HEAD FENCE IS A DELAY-SLOT THREAD SELECTOR, AND IT ONLY FIRES WHEN `mostly_true_jump > 0` (amendment to §164-36; its "−1 instruction" tell is falsified) L21030 - **§204-D** — A LOOP-INVARIANT LOAD IS ADMITTED AS A MOVABLE ONLY IF ITS ADDRESS CANNOT TRAP: `local.field` PASSES, THE SAME READ THROUGH A POINTER LOCAL IS REFUSED L21814 - **§225** — THREE CONTROL-FLOW SHAPES NO STRUCTURED SPELLING REACHES (P31 S58) L22985 - **§30** — addendum (P31 S58) — the `/s` grant closes a SCHEDULING residual and a REGISTER residual with one edit L23267 - **§249** — THE SELF-ASSIGN, THE DEAD RE-ASSIGN, AND THE `+ zr` COPY: THREE WAYS TO MAKE A DELETED INSTRUCTION REAL (P31 S58b) L23941 - **§30** — addendum (P31 S58b) — THE ANONYMOUS STRUCT MEMBER REF GRANTS `/s`, AND THAT IS A TWO-FOR-ONE L24209 - **§263** — A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCHEDULE (P31 S59, byte-proven) L24937 - **ADD-8** — → §245 addendum — THE RE-TIE BARRIER PINS CALL-ARG SETUP TO SOURCE ORDER (construct 6) L25213 - **ADD-11** — → Cross-confirmation card block (per §259's standing instruction: confirmation, not news) L25244 - **ADD-4** — → §167-37 addendum — THE FIFTH PRECONDITION, NOW MEASURED: DOWNSTREAM FIELD RE-READS ⇒ NAME IT **AND** STORE **INSIDE** THE ARM L25404 - **ADD-10** — → Cross-confirmation card block (per §259's standing instruction: confirmation, not news) L25495 - **ADDENDUM** — to §82 — struct copies must stay MEMBER-WISE, not block-moved, to match a spill-slot save L25753 - **ADDENDUM** — to §244 — THE COMPILER BARRIER, NOT `volatile`, BLOCKS A STORE→LOAD HOIST THROUGH A POINTER-OFFSET FIELD L26079 - **§281** — GROUP COPY-THEN-RMW BY OPERATION KIND, NOT FIELD BY FIELD: sched1 does the interleaving, the source must not (P31 S60; wave cf, func_80180FB4, ov_SC03_111, byte-proven) L27000 - **§NNN** — A DEPENDENT COPY-THEN-RMW BLOCK MUST BE SOURCE-GROUPED BY OPERATION KIND, NOT BY FIELD, AND SCHED1 DOES THE INTERLEAVING (P31 S60; `func_80180FB4`, ov_SC03_111, byte-proven) L27001 - **§NNN** — WRITE THE UP-COUNT LOOP: gcc's OWN REVERSAL PRODUCES A COUNTER-INIT INSTRUCTION THAT LANDS AFTER HOISTED MOVABLES, WHICH A HAND-WRITTEN DOWN-COUNT LOOP CANNOT REPRODUCE (P31 S60; `func_80180DD8`, ov_SC04_005, byte-proven) L27047 - **§285** — PRE-INITIALIZING A VARIABLE WITH A SHARED CONSTANT BEFORE A BRANCH MAKES BOTH ARMS OF THE FOLLOWING if/else DESTRUCTIVELY REUSE THE SAME DESTINATION REGISTER FOR THEIR BITWISE RESULT (byte-proven; `func_801811F0`, ov_SC03_102, independently rediscovered across waves di/dj) (P31 S60; waves #, byte-proven) L29088 - **§287** — A STACK-FRAME HOLE BELOW TWO ADDRESS-TAKEN AGGREGATE LOCALS IS A LEADING PAD MEMBER OF ONE COMBINED STRUCT, NOT SEPARATE LOCALS OR A REGISTER PIN (P31 S60/dj; `func_8017D104`, ov_SC06_027, byte-proven 47/47) (P31 S60; waves #, byte-proven) L29096 - **§297** — ONE GIV SERVING MIXED-WIDTH LOADS *AND* A STORE NEEDS STRUCT-MEMBER SPELLING; CAST/INDEX SPELLINGS RE-SPLIT THE PSEUDO (P31 S61; wave m0a, `func_800347C8`, main, byte-proven 31/31) L29509 - **§308a** — A TWO-INSTRUCTION ARM STUB LAID *BEFORE* THE OTHER ARM AND REACHED ONLY BY A TRAILING `j` SURVIVES ONLY BEHIND A SHORT-CIRCUIT `&&` GUARD PLUS A SEPARATE RE-TEST; EVERY PLAIN if/else-if, GOTO-LADDER, NESTED-INVERTED-if AND 1-/2-NODE switch LETS jump1 INLINE IT (−2 ins) (P31 S63 t5e-t5i; byte-proven func_80180808) L30022 - **§312** — A BARE RELATIONAL IN AN `if` GIVES THE `slt` NO TARGET: ASSIGN THE COMPARISON INTO THE ALREADY-ALLOCATED VARIABLE TO CHOOSE ITS DESTINATION REGISTER, AT ZERO INSTRUCTION COST (P31 S64 t5o-t5r; byte-proven func_80180DCC) L30169 - **§314** — AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES ALL COMPILE BYTE-IDENTICAL, AND NONE REACHES A TARGET THAT SINKS THE NEGATIVE ARM INTO AN OUT-OF-LINE BLOCK (P31 S65; UNPROVEN, gate-refused draft func_80181720) L30251 - **§315** — ALL-CONSTANT AGGREGATE FILL: THE EMISSION ORDER IS SHARED-LITERAL GROUPS × DESCENDING INDEX, WITH PAIRED SUB-FIELDS INTERLEAVED (P31 S66; ⚠ **UNPROVEN** — `func_80185214`, gate-REFUSED draft, closeness 32→5, never MATCHed) L30283 - **§316** — A GUARD CHAIN THAT GATES A SUCCESS BLOCK ROUTES ITS FAILING GUARDS TO A LABEL **BEFORE** THE FLAG TEST, NOT TO THE EPILOGUE: THE RESIDUAL IS IMM-OFFSET-ONLY AT AN EXACTLY EQUAL INSTRUCTION COUNT (P31 S66; byte-proven func_80180284) L30320 - **§317** — A NARROW STRUCT-FIELD STORE TURNS A *LIVE SImode VARIABLE* INTO A `li`: convert.c's TRUNCATION-DISTRIBUTION MINTS A HImode COPY, AND cse REWRITES THAT COPY'S SOURCE TO THE EQUIVALENT CONSTANT (P31 S66; byte-proven func_8017EF94) L30354 - **Addendum** — WIDTH sh!=lhu residual: hoist a sibling field's load first (func_80182538) L30582 - **Addendum** — Masked-OR field-merge plateaus at close=10; bitfield store clears it (func_8017FD64) L30710 - **§321** — FILE-SCOPE DUPLICATE ANONYMOUS-STRUCT TYPEDEFS ARE A HARD ERROR; THE SAME TEXT AT BLOCK SCOPE IS A WARNING (P31 S66; byte-proven func_80180728, func_8017F9F8, func_8017E07C) L30855 - **§335** — AN `extern u16 A[]` READ AT A VARIABLE SUBSCRIPT ALLOCATES DEAD STACK TEMPS (~8B PER ACCESS) THAT INFLATE THE FRAME WHILE EMITTING ZERO EXTRA INSTRUCTIONS (P31 S67; byte-proven ov_SC06_025/func_8017EA74, closeness 141 -> 20) L31150 - **§346** — `c ? X : -X` TAKES expand_expr's COND_EXPR **SINGLETON** PATH (copy, then negate IN PLACE) — AN if/else STATEMENT GIVES THE TWO-ARM FORM (P31 S67; byte-proven ov_SC03_102/func_80180C38, closed the last instruction) L31332 - **§351** — `/s` (MEM_IN_STRUCT_P) IS A DIAL YOU CHOOSE PER ACCESS: A COMPONENT_REF GRANTS IT AND LETS cse KEEP AN INDEX ACROSS THE STORE; A PLAIN CAST DENIES IT (P31 S67; ov_SC01_000/func_8017DD04, 186 -> 5) L31466 - **§357** — ONE STRUCT POINTER, NOT TWO: A SECOND SOURCE VARIABLE BUILDS A THIRD IV (P31 S68; byte-proven ov_SC06_029/func_80181DF8, 335 ins, 330 → 13) L31599 - **§364** — ★ — THE libgpu `P_TAG` BITFIELD SPELLING IS **OPT-LEVEL DEPENDENT** (P31 S68; two functions, opposite verdicts, same session) L31744 - **§379** — ★★★ — **MEM_IN_STRUCT_P**: THE SAME LOAD, WRITTEN AS A STRUCT MEMBER, SCHEDULES WHERE A CAST CANNOT (P31 S69; byte-proven main/func_80021284 220 ins and main/func_8002D904 217 ins, found INDEPENDENTLY by two agents) L32174 - **§391** — ★★ — A BYTE-ALIGNED STRUCT COPIES IN FOUR INSTRUCTIONS, A WORD-ALIGNED ONE IN TWO (P31 S69) L32498 - **§395** — ★★★ — FIVE NARROWING/PLACEMENT LEVERS FROM ONE 91-INSTRUCTION CRACK (P31 S69; byte-proven ov_SC06_018/func_80189E60, warm start 32 off → MATCH 91/91) L32621 - **§418** — ★★★ — TWO LOOP-STRUCTURE LEVERS: MAKE THE SECOND INDEX A GIV, AND KEEP A TABLE ADDRESS UNFOLDED (P31 S71; byte-proven `ov_SC04_016/func_8017DF8C`, 184 ins, 32 → 0 in seven compiles) L33642 - **§425** — ★★★ — `sb` ALIASES SCALAR GLOBALS WHILE `sh`/`sw` STRUCT STORES DO NOT, AND TWO MORE ALIAS/BOOST RULES (P31 S71; `md_MAIN_003/func_800CF3E8`, 467 of 469 ins, all four byte-verified from `-dS`/`-dR`/`-dl`/`-dr`) L33839 - **§442** — ★★★ — A RECOVERY RUNG THAT REWRITES A LEGAL CONSTRUCT INTO AN ILLEGAL ONE READS EXACTLY LIKE A CODEGEN WALL (P31 S75; `reconcile_tu`, 344 ins unblocked) L34657 - **§468** — THE `%lo`-FOLD EXTENDS TO STORES ONLY VIA `extern Struct SYM[]`, AND MASKING HID THE OPERAND ORDER L35390 - **§469** — THE `MEM_IN_STRUCT_P` ALIAS UNLOCK (and §463's spill law, independently confirmed) L35424 - **§497** — ★ — A BODILESS `typedef struct Tag Alias;` DEFINES THE ALIAS, NOT THE TAG: THE CARRIER'S FALSE "CONFLICTING BODIES" REFUSAL (P32 T1b; ov_SC02_017 `func_80186C64` isolation) L36705 - **§458** — Struct spelling is a per-access dial: the T2 probe's byte evidence (Phase 37 S106) L37886 ### types, signedness & load/store width (98) - **§3-I1** — Unsigned range check: `(x - lo) < (hi-lo)` → `addiu`+`sltiu` L41 - **§3-I2** — Byte mask forces `andi` even after `lbu` L47 - **§3-T3** — Types L85 - **§7** — PsyQ SDK types & symbols (the library-call prerequisite) L322 - **§14c** — Callee-signature-aware harvest — breaking the extern-type-conflict wall (Phase 15, T6) L1236 - **Phase-20** — RESOLUTION — `tools/cast_call_sites.py` BUILT + the cap re-diagnosed (R14, byte-proven) L1724 - **§23** — Giant `func_80153E00` cracked (scalar-data CAST); BUT most giant drafts are STALE+INCOMPLETE, and a diag MUST remove artifacts (Phase 21 — byte-proven + a self-correction) L2055 - **§28b** — The struct-walled close=0 giant (§28 case #3) IS bankable: the engine_types.h type-lift (Phase 22 T2, byte-proven on func_80156B74 ×134) L2380 - **§33** — Automating the giant decl-reconcile: `tools/reconcile_decls.py` (the DATA analog of §20's `cast_call_sites`) + a fleet-majority type oracle (Phase 24 T7b, byte-proven on `func_80129CF8`) L2457 - **§40a** — The DECL-RECONCILE pass (type-lift so remapped drafts compile in the sibling TU) (Phase 25 T7.2, 2026-07-08) L2624 - **§41** — The DEF-SIDE canonical-sig wall: mechanically banking a drafted giant past `conflicting types` (Phase 25 T5b batch-2, 2026-07-09; `tools/canon_sig_reconcile.py`, byte-proven on `func_8013B274`) L2733 - **§41d** — `void`→`s32` is NOT always byte-neutral: gate the RAW draft FIRST (Phase 26 session 6, byte-proven) L2916 - **§42b** — addendum — wave 3 (Max, 2026-07-10c): THE STALE-OBJECT GATE TRAP + the read-global `&`-cast drift + fix L3032 - **§43** — The K&R s16-param definition DISSOLVES the "narrow-param wall" for by-value register args (Phase 25 task A, Fable5 crack of the 369-ins giant `func_80166994` ×134, 2026-07-11) L3233 - **§3-C.** — TYPE- AND SHAPE-DRIVEN CODEGEN (the C type literally selects the addressing mode) L3522 - **§56b** — PROPAGATING an h_seq giant: the exemplar's externs MUST be fleet-canonical, not the draft's types L4270 - **§64** — The §20 type-lift's three laws: fold the tagged typedef, check VISIBILITY, and strip only what is TEXTUALLY IDENTICAL (`lift_types.py`, Phase 29 SESSION-14, 2026-07-23) L4946 - **§64a** — VARIANT types: UNIQUIFY the camps, do not reconcile them (`uniquify_type.py`, Phase 29 SESSION-14) L4995 - **§65g** — Where the cheap levers STOP: the local-type and in-TU-self-decl classes did not yield L5177 - **§68** — A comment-only line halted the extern scan, and the skip label blamed the type cap (Phase 29 SESSION-18) L5562 - **§73** — A def-side self-decl conflict has TWO axes: RETURN (fleet widen, T2) and PARAMS (casts at each use, T0). Diagnose which before paying for the expensive one (Phase 29 SESSION-19, `func_8014F3E8` + `func_8014D4C0`) L5821 - **§75c** — Class-B's remedy is the FULL §17a-1 PAIR (decl **and** call-site cast); a decl-only fix moves the error and looks like a new wall (Phase 29 SESSION-19, `func_8012F14C`) L6035 - **§3-The** — defect this exposed: a shared type that is present but invisible L6392 - **§92** — Conforming a DECLARATION: pointer changes are caller-neutral, scalar-WIDTH changes are not (Phase 29 SESSION-21, `tools/conform_decls.py`) L6929 - **§94** — A family sweep's `0/N` is a TYPE-CARRY failure until proven otherwise: lift the exemplar's local types, ALL of them, transitively (Phase 29 SESSION-21, `func_8016B6BC` 0/137 → 137/137) L6985 - **§100** — Prefer the DRAFT-LOCAL fix: a type only one function uses belongs in its BODY, not in a shared header (Phase 29 SESSION-22, `func_80175DA8` 0/137 → 137/137) L7254 - **§120** — Uniquify draft-defined TYPE names; and check which of N staging sites you actually patched (Phase 29 T93) L8064 - **§3-Do** — NOT "strip the duplicate typedef" — it breaks the extern that uses it L8075 - **§3-The** — type-form rules L8917 - **§143** — `cast_call_sites` read a RETURN STATEMENT as a prototype and deleted it. A 0/39 sweep became 18/39. (P30 S40) L9872 - **Then** — propagation returned 0/137 TWICE — both times a missing TYPE L10038 - **§154** — Reading a disc payload: the module-id word, static base derivation, and "type 1 = uncompressed overlay" (P30 S44) L10560 - **§3-C.** — PAC type 1 = the same payload class as type 4, just NOT compressed L10586 - **§155b** — check the TYPE your oracle returns before comparing against it (S45 p5) L10625 - **§3-B.** — Typedef handling — the only strategy that survives contact L17062 - **§3-D.** — A NARROW TYPE BLOCKS COPY ELISION (func_8001D3FC — new idiom) L17333 - **§185** — EDIT THE SIDE THAT IS CHEAP TO VERIFY, AND CHECK A TU RETYPE AT ITS USE SITES L18105 - **§186b** — A NO-SAVE 16-BYTE FRAME IN A LEAF FUNCTION MEANS `s16` LOCALS, NOT A HIDDEN CALL L18158 - **§193-B** — A NARROW CAST OF A WIDE PARAMETER COSTS ONE EXTRA INSTRUCTION WHEN ITS STATEMENT SITS AFTER AN INTERVENING `jal` — AND THE DECIDER IS `combine.c:929`'s CROSS-CALL GUARD, NOT REGISTER ALLOCATION L18554 - **§193-G** — §164-54's "scope to ≥4 arms" bound is byte-wrong — the dispatch-topology oracle goes live at THREE case nodes (`balance_case_nodes` splits at `i > 2`), but only for a signed-after-promotion index L18791 - **§193-H** — A pointer-derived base load `*(s32*)(p+K)` is uncacheable across ANY memory write or call (cse's third kill disjunct) — so the target emits one `lw` per store-separated RUN of spellings, and a run of stores sharing one `lw` is a C local you must declare L18852 - **§194-B** — A `addu $rA,$rB,$zero` copy feeding ≥2 `sh` stores is a SECOND, 16-BIT-DECLARED local — the width, not the clamp or the join liveness, is what keeps the copy (and func_80183094's pin + `"memory"` clobber are both provably inert) L19061 - **§194-D** — Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication + attribution correction of §165-17, NOT a new argument-register law L19162 - **§194-G** — Reading the integer half of a 16.16 stack aggregate: REGISTER-LIVENESS, not the C spelling, decides `sra` vs `lhu` — and the break is count-neutral L19313 - **§194-H** — §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a branch fences the store that reads v, and when TWO independent stores compete for the one delay slot the fence picks the winner at ZERO length drift (a WIDTH/sh!=sw swap, not a nop). Corrects §167-42's reconstruction (pseudo arm → hard-reg arm) and supplies its first re-runnable A/B. L19356 - **§194-I** — §16N+2's magic-per-odd-part ladder has exactly one broken row — read the divisor arithmetically instead: d = round(2^(32 + post_shift) / magic_read_as_unsigned) L19394 - **§197-A** — A NARROW SIGNED MEMORY READ FEEDING A CONSTANT `>>` LOSES ITS `lh`, AND THE CURE IS AN ASM RE-TIE (attribution CONTESTED: cse vs combine) L20626 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§201-D** — THE SIGNEDNESS OF A div/mod MAGIC IS DECIDED BY THE STATIC TYPE OF THE DIVIDEND TREE AFTER `get_narrower` STRIPS WIDENING CONVERSIONS — NEVER BY A PROVABLE RANGE. AN `& 0xFF` IS NOT A CONVERSION, SO IT NEVER FLIPS THE MAGIC; A DECLARED-UNSIGNED LOCAL *OR* A NARROWING CAST WRITTEN AT THE DIVIDE BOTH DO. L21372 - **§203** — A DEDUPED TYPEDEF MUST PRECEDE EVERY SPLICE POINT, NOT JUST ITS OWN (P31 S56) L21537 - **§204-E** — `decl_prior`'s `%hi/%lo` ARM HAS NEVER FIRED: the card's promised GLOBAL-TYPE row is 0 of 1,210 L21889 - **§218** — A NARROW TYPE AT THE ABI BOUNDARY COSTS AN IN-PLACE `sll/sra` PAIR — on the RETURN as well as on the PARAMETER (P31 S58) L22733 - **§220** — THE PARAMETER ITSELF IS A REGALLOC DIAL: use it directly, prefer `s32` to `void*`, and place the save-copy AFTER the first call (P31 S58) L22789 - **§227** — TYPE THE SOURCE BY THE **LOAD** WIDTH, NOT BY THE STORE WIDTH (P31 S58) L23069 - **§234** — CONSTANT MATERIALISATION: THE STORE LVALUE'S SIGNEDNESS PICKS `addiu` vs `li`/`ori` (P31 S58b) L23367 - **§237** — THE CAST-AT-CALL-SITE DECISION TABLE: WHAT §17a-1 FIXES, WHAT IT CANNOT, AND THE FOUR ESCAPES (P31 S58b) L23516 - **§242** — `*k` vs `<>n`: EXPRESSION SPELLING OWNS THE LOAD WIDTH AND THE ROUNDING CHAIN (P31 S58b) L23703 - **§243** — SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P31 S58b) L23731 - **§251** — IMMEDIATE-SPELLING TRIGGERS: `+= 0xFF`, FULL-WIDTH `~K`, AND THE TWO-OR SPLIT (P31 S58b) L24020 - **ADD-5** — → §1/I1 addendum — THE INVERTED RANGE TEST: `(u32)(x-lo) >= N` WITH THE ZERO-ARM AS THE TRAILING `else` L25183 - **ADD-2** — → §236 addendum (item 1 corollary) — THE SILENT-SPLICE DECLARATION LADDER: WHERE THE TU HAS NO SPELLING, UNPROTOTYPED `()` BEATS THE FLEET VOTE L25377 - **ADD-5** — → §267-ADD-6 / §172b-1 addendum — THE SIGN-TEST FACE: `(s16)v < 0` IS `sll 16` + `bgez` ON THE SHIFTED COPY, RAW STAYS LIVE — AND THE HALFWORD-ABS SHAPE NEEDS NO RITUAL L25421 - **ADD-8** — → §30a addendum — THE THIRD COLUMN: `*p++` IS ALSO SERIALIZED BUT BURNS `addiu`; ONLY CAST-PLUS GIVES SERIALIZATION *AND* DISPLACEMENT FOLDING L25466 - **ADDENDUM** — to §179-C — the `.type NAME, @function` requirement L25650 - **ADDENDUM** — to §134 — a typedef defined BELOW the splice point is stripped anyway L25697 - **ADDENDUM** — to §172b-1 — TWO PLACEMENT DIALS THE PROMOTION LAW DOESN'T NAME: NARROW THE COUNTER'S DECLARED TYPE, AND MOVE THE CAST INTO THE LOOP BODY L26110 - **ADDENDUM** — to §172b-4 — THE PLAIN CAST-DIVISION ALREADY PRODUCES THE PATTERN; DON'T HAND-ROLL THE BIAS, AND KEEP THE OPERAND WIDE L26202 - **§280** — THE CURSOR'S DECLARED TYPE PICKS `sltu` vs `slt`: a `T*` bound test is UNSIGNED by C rule, an integer cursor is signed (P31 S60; wave cf, func_800CAE74, md_MAIN_031, byte-proven) L26966 - **§NNN** — A LOOP CURSOR'S C TYPE (POINTER vs PLAIN INTEGER) SELECTS `sltu` vs `slt` FOR ITS BOUND TEST, INDEPENDENT OF THE VALUES INVOLVED (P31 S60; `func_800CAE74`, md_MAIN_031, byte-proven; cross-confirmed same wave by `func_8017F2A4`, ov_SC03_096) L26967 - **ADDENDUM** — to §21 — the `bltz`+`slti` (or N-separate-compares) signed-range-split bullet is now CONFIRMED on three independent functions, and generalizes beyond `lbu`/u8 L27167 - **ADDENDUM** — to §202 — the DEF-SIDE ALIAS also resolves a function-vs-DATA-symbol identifier clash, not only a function-vs-function prototype clash L27213 - **ADDENDUM** — §NNN — sharpens §162a3/§60b: A `sll $v0,16 / sltiu $v0,1` ZERO-TEST OF A JUST-DECREMENTED HALFWORD IS A SIGNED-TEMP WIDTH TELL OUTSIDE ANY SWITCH/RANGE-TEST CONTEXT L28303 - **§288** — A REGISTER PIN DECLARED UNINITIALIZED AND ASSIGNED ONLY AT ITS LATE, SOLE USE STILL FORCES THE FIXED REGISTER'S SAVE/RESTORE, AND THE ASSIGNMENT'S SOURCE POSITION CONTROLS WHERE THE VALUE MATERIALIZES (P31; `func_80186530`, ov_SC02_017, byte-proven, match_one MATCH re-verified) (P31 S60; waves #, byte-proven) L29100 - **§292** — DECLARING A SYMBOL UPSTREAM OF AN ALREADY-BANKED SIBLING THAT RELIES ON THAT SYMBOL'S IMPLICIT (K&R) DECLARATION CAN SILENTLY REPROTOTYPE THE SIBLING'S OWN CALL SITE (P31 S60; waves #, byte-proven) L29116 - **ADDENDUM** — to §172b-1 (counter-type dial) — the sll's SOURCE REGISTER is the placement discriminator L29278 - **ADDENDUM** — to §31's asm→layout inference — TWO WIDTH-PAIR DISCRIMINATORS (same offset, different widths) L29390 - **§297** — ONE GIV SERVING MIXED-WIDTH LOADS *AND* A STORE NEEDS STRUCT-MEMBER SPELLING; CAST/INDEX SPELLINGS RE-SPLIT THE PSEUDO (P31 S61; wave m0a, `func_800347C8`, main, byte-proven 31/31) L29509 - **§318** — A UNARY MINUS STORED BACK INTO THE SAME HALFWORD IS COMPUTED IN HImode, SO ITS LOAD IS `lhu`; AN INLINE `(s32)` CAST IS PROVABLY INERT AND ONLY A NAMED SImode LOCAL BUYS `lh` (P31 S66; `func_8018568C`, ov_SC03_028 — **`match_one`-only, NOT byte-gated**) L30388 - **Addendum** — Reversed-operand sltu then xori 1 means unsigned less-or-equal (func_8017D948) L30515 - **Addendum** — WIDTH sh!=lhu residual: hoist a sibling field's load first (func_80182538) L30582 - **Addendum** — Static local_type blocker survives typedef removal — it's textual (func_801588CC) L30625 - **§320** — THE §43 "RETURN-TYPE FLIP PAIR" IS **NOT** TU-EDIT-REQUIRED: THREE DRAFT-ONLY ESCAPES (P31 S66; byte-proven func_800CCBC0, func_800D30D0, func_800D2A24) L30796 - **§321** — FILE-SCOPE DUPLICATE ANONYMOUS-STRUCT TYPEDEFS ARE A HARD ERROR; THE SAME TEXT AT BLOCK SCOPE IS A WARNING (P31 S66; byte-proven func_80180728, func_8017F9F8, func_8017E07C) L30855 - **§323** — CARRYING FILE-SCOPE TYPES BETWEEN SPLIT TUs: DEDUPE BY NAME, SKIP HEADER-PROVIDED, AND THE GUARD TRAVELS WITH THE BLOCK (P31 S67; ov_SC02_000, two defects fixed, one open) L30906 - **§327** — A RANGE TEST MUST BE HImode: WITH `s32` + A `(u16)` CAST GCC *PROVES* THE MASK REDUNDANT AND DROPS THE `andi` (P31 S67; byte-proven ov_SC02_039/func_8017EC34, 154 ins) L31019 - **§328** — THE VOLATILE ALIAS MUST BE AN *OBJECT*, NOT A CAST: `*(volatile s32*)&sym` UNFOLDS %lo INTO A SEPARATE `addiu` (P31 S67; byte-proven ov_SC07_007/func_80181B8C, 156 ins, NEW LAW) L31034 - **§335** — AN `extern u16 A[]` READ AT A VARIABLE SUBSCRIPT ALLOCATES DEAD STACK TEMPS (~8B PER ACCESS) THAT INFLATE THE FRAME WHILE EMITTING ZERO EXTRA INSTRUCTIONS (P31 S67; byte-proven ov_SC06_025/func_8017EA74, closeness 141 -> 20) L31150 - **§342** — A `void *` PARAMETER CAST TO ITS REAL TYPE IN A LOCAL IS **NOT** BYTE-NEUTRAL WHEN A LATER PARAMETER ALSO NEEDS A CALLEE-SAVED REGISTER (P31 S67; byte-proven ov_SC07_002/func_80181394, NEW LAW) L31267 - **§345** — A VOLATILE **STORE** EVICTS THE MEM FROM cse AND KEEPS `sh`; A VOLATILE **LOAD** BLOCKS combine AND DEGRADES `lh` INTO `lhu+sll+sra` (P31 S67; byte-proven ov_SC01_084/func_80181A7C) L31322 - **§351** — `/s` (MEM_IN_STRUCT_P) IS A DIAL YOU CHOOSE PER ACCESS: A COMPONENT_REF GRANTS IT AND LETS cse KEEP AN INDEX ACROSS THE STORE; A PLAIN CAST DENIES IT (P31 S67; ov_SC01_000/func_8017DD04, 186 -> 5) L31466 - **§378** — ★★★ — THE **SELF-CALLER CAST**: LET A TU KEEP CALLING THE FUNCTION IT IS ABOUT TO DEFINE (P31 S69; byte-proven ov_SC04_010/func_8017D6CC) L32124 - **§379** — ★★★ — **MEM_IN_STRUCT_P**: THE SAME LOAD, WRITTEN AS A STRUCT MEMBER, SCHEDULES WHERE A CAST CANNOT (P31 S69; byte-proven main/func_80021284 220 ins and main/func_8002D904 217 ins, found INDEPENDENTLY by two agents) L32174 - **§392** — ★★★ — THE SONNET-WAVE HARVEST (P31 S69): SEVEN TYPE/ORDER LEVERS THAT EACH CLOSED A MATCH L32512 - **§422** — ★★ — QImode ARITHMETIC VIA `(u8)(x - K)`, AND `flag ^ 1` NEEDS ITS OWN TEMP (P31 S71; byte-proven `resident/func_800D06E8`, 344 ins) L33770 - **§423** — ★★★ — "MATCH IN ISOLATION + GATE REJECTS + CAUSE NOT DETERMINED" ⇒ GREP THE TU FOR A FILE-SCOPE TYPEDEF THE DRAFT ALSO CARRIES (P31 S71; byte-proven `ov_SC03_092/func_8017FA74`) L33790 - **§481** — ★★★ — `conflicting types` IS A SAME-SCOPE ERROR; ACROSS SCOPES IT IS ONLY A WARNING (P31 S77, `main:func_8001FC08`, 400 ins) L35911 - **§496** — ★★ — A CARRIED-TYPE TEST THAT ASSUMES THE OVERLAY INCLUDE SET SILENTLY DROPS A RESIDENT TYPEDEF (P32 T1a; byte-proven, resident `func_800D128C` isolation) L36671 - **§497** — ★ — A BODILESS `typedef struct Tag Alias;` DEFINES THE ALIAS, NOT THE TAG: THE CARRIER'S FALSE "CONFLICTING BODIES" REFUSAL (P32 T1b; ov_SC02_017 `func_80186C64` isolation) L36705 ### declarations, prototypes & K&R (122) - **§3-T4** — Branch polarity: invert the source condition to flip gcc's chosen branch L90 - **§8c** — Splitting a TU means rebuilding its DECLARATION ENVIRONMENT, not moving text (Phase 26 session 6) L456 - **§8d** — Templating a body INTO a TU must not CHANGE its declaration environment — demote the carried data externs (Phase 26 session 8, byte-proven on `func_8015AE2C` ×133) L502 - **§13** — Add a location overlay — the canonical runbook (Phase 13; the Phase-15 fleet recipe) L1075 - **§14c** — Callee-signature-aware harvest — breaking the extern-type-conflict wall (Phase 15, T6) L1236 - **§3-The** — pipeline gotcha — match_one ≠ the gate; `sig_unify` is MANDATORY L1444 - **§3-The** — gate two-stage — sig_unify is a FALLBACK, not unconditional (`gate_stage.py`, §19 folded in) L2212 - **§28** — Banking a "close=0 gate-rejected" giant: the canonical-extern recovery (Phase 22 T1, byte-proven on func_8015126C ×134) L2348 - **§28d** — The "macro-extern-injection" lever: freeing reach-134 inline matches dedup_propagate skips as "not self-contained" (Phase 23, `tools/inject_capped_externs.py`) L2389 - **§33** — Automating the giant decl-reconcile: `tools/reconcile_decls.py` (the DATA analog of §20's `cast_call_sites`) + a fleet-majority type oracle (Phase 24 T7b, byte-proven on `func_80129CF8`) L2457 - **§40a** — The DECL-RECONCILE pass (type-lift so remapped drafts compile in the sibling TU) (Phase 25 T7.2, 2026-07-08) L2624 - **§41** — The DEF-SIDE canonical-sig wall: mechanically banking a drafted giant past `conflicting types` (Phase 25 T5b batch-2, 2026-07-09; `tools/canon_sig_reconcile.py`, byte-proven on `func_8013B274`) L2733 - **§41b** — addendum — M4 "reconcile_decls" tier is ALSO a probe over-count: 0/8 mechanical (Phase 25 T7-M4, 2026-07-10) L2896 - **§41d** — `void`→`s32` is NOT always byte-neutral: gate the RAW draft FIRST (Phase 26 session 6, byte-proven) L2916 - **§43** — The K&R s16-param definition DISSOLVES the "narrow-param wall" for by-value register args (Phase 25 task A, Fable5 crack of the 369-ins giant `func_80166994` ×134, 2026-07-11) L3233 - **§51g** — When the thing you are scanning has a GRAMMAR, parse the grammar (`tools/cdecl.py`) L3807 - **§54** — `--fix-def-sig`: the member's CANONICAL DECLARATION is a build step too (tiny-IMM mega-pools, +4,801, Phase 29 T6, 2026-07-16) L4201 - **§56** — Banking a hand-drafted GIANT into its exemplar TU: self-contained draft vs live-TU decls, 4 reconciliation tactics (`func_8013FAF8`, 312 ins ×138, Phase 29 T4, 2026-07-17) L4225 - **§56b** — PROPAGATING an h_seq giant: the exemplar's externs MUST be fleet-canonical, not the draft's types L4270 - **§57** — The SELF-decl normalize: the sibling's OWN caller declares the templated fn divergently (the third §17a-1 direction; `tools/normalize_self_decls.py`, func_801670E4 ×137, Phase 29, 2026-07-18) L4294 - **§61** — Task 14: the gate ladder's missing stage is the ARITY pre-pass, and it is TU-side not draft-side (Phase 29, 2026-07-21) L4583 - **§63** — The fresh-138 DEF-SIDE blocker: fix the HEADER decl, not the draft (`fix_header_decl.py`, Phase 29 SESSION-13, 2026-07-23) L4915 - **§63** — UPDATE (Phase 29 SESSION-14) — `fix_header_decl`'s "SAFE" verdict is FLEET-BLIND; it MUST be R22-validated L5038 - **§65c** — `rtu_match` MATCH → bank held 13/13 on self-decl, but broke on the FIRST callee-decl case L5119 - **§65g** — Where the cheap levers STOP: the local-type and in-TU-self-decl classes did not yield L5177 - **§66a** — The widest write in a pipeline is the one most likely to be UNDECLARED L5219 - **§68** — A comment-only line halted the extern scan, and the skip label blamed the type cap (Phase 29 SESSION-18) L5562 - **§73** — A def-side self-decl conflict has TWO axes: RETURN (fleet widen, T2) and PARAMS (casts at each use, T0). Diagnose which before paying for the expensive one (Phase 29 SESSION-19, `func_8014F3E8` + `func_8014D4C0`) L5821 - **§75** — A propagation cap is usually a MINORITY-SPELLING SOURCE OVERLAY, not a wall: census the carried extern before believing the exclusion message (Phase 29 SESSION-19, `func_8014F3E8` ×4 → ×138) L5904 - **§75b** — A body's preamble can carry `#define`s, not just `extern`s; extraction lifts only the externs (Phase 29 SESSION-19, `func_80165CA0` ×3 → fleet) L5986 - **§75c** — Class-B's remedy is the FULL §17a-1 PAIR (decl **and** call-site cast); a decl-only fix moves the error and looks like a new wall (Phase 29 SESSION-19, `func_8012F14C`) L6035 - **§76** — The allocno CLASS (local vs global) is the dominant regalloc lever, and C reaches it ONLY through declaration scope and variable reuse (Phase 29 SESSION-19, behemoth #3 `func_8017F510` 1,511 ins, 97 → MATCH, pin-free) L6066 - **§79** — For a 0-callee giant, fingerprint by DATA symbols (§71 cannot fire); and the STACK-SLOT ORDER is a declaration-order oracle (Phase 29 SESSION-19, `func_8017BF14` 4,763 ins, cold start → 45/4763) L6258 - **§3-NEW** — LEVER — the frame layout reads back the original declaration order L6278 - **§3-1.** — The inlined-helper signature L6413 - **§92** — Conforming a DECLARATION: pointer changes are caller-neutral, scalar-WIDTH changes are not (Phase 29 SESSION-21, `tools/conform_decls.py`) L6929 - **§95** — `reconcile_tu` dropped the SIBLING declarators of a multi-symbol `extern` line (Phase 29 SESSION-21, `func_80176218`) L7021 - **§96** — The same rewrite, one shape down: `reconcile_tu` matched statements to lines by TEXT, so every COMMENTED declaration was silently skipped (Phase 29 SESSION-22, `func_80176218` banked) L7056 - **§98** — `conform_decls` had three defects, and only the third needed R22 to find (Phase 29 SESSION-22, `func_8014CF04`) L7151 - **§99** — The narrow-param wall is a DEF-side problem with a ZERO-blast-radius fix: convert the definition to K&R (Phase 29 SESSION-22, `func_80175AB8` + `func_80175DA8`) L7205 - **§102** — A PLUMBING verdict can MASK a DIFF; and K&R is not always a codegen change (Phase 29 SESSION-22, `func_8016EC0C`) L7310 - **§103** — A FILE-scope `extern` in a shared overlay TU is a GLOBAL constraint on every LATER function; move the DECL, not the draft (Phase 29 T48/T51, `func_80135260` — the fleet-wide half) L7345 - **§104** — Two silent-skip defects in one scan: match on MASKED text, emit from the ORIGINAL (Phase 29 T53, `gather_externs`) L7433 - **§112** — A macro-scoped declaration only collides where the macro is INSTANTIATED (Phase 29 T67/T69, `audit_header_sigs.py`) L7768 - **§113** — An ARITY blocker only exists if the macro CALLS the function; an address-taken use has no call site (Phase 29 T72) L7814 - **§114** — The THIRD decl axis: a CALLEE the draft declares differently from the target TU (Phase 29 T76/T77) L7846 - **§3-Do** — NOT "strip the duplicate typedef" — it breaks the extern that uses it L8075 - **§121** — Synthesise externs for macro-DEFINED callees from the macro's own definition head (Phase 29 T95) L8099 - **§3-The** — declaration surface (integration, not codegen) L8979 - **§136f** — Two declaration sub-cases the reconcile lane surfaced (lane now 15/15 lifetime) L9224 - **§138** — The propagation lanes: a gate refusal is a DECLARATION, and which lever you owe depends on blast radius L9453 - **Reconciling** — a gate-refused draft: which way you edit depends on WHERE the TU's decl is L9603 - **§141** — The §134 class is CLOSED: every line-shape decision now routes through `cdecl._mask` (P30 S39) L9778 - **§143** — `cast_call_sites` read a RETURN STATEMENT as a prototype and deleted it. A 0/39 sweep became 18/39. (P30 S40) L9872 - **§3-C.** — Inner-block declaration does NOT delay slot allocation — BYTE-REFUTED L10106 - **§159** — THE DECLARATION AXIS: conform to byte-truth, and make every guard state its COVERAGE (P30 S47; ~10,930 sites across 8 axes, fleet byte-identical) L10866 - **§168** — THE COUSIN TIER (P30 S49, 2026-08-12): h_seq's exact-hash brittleness, measured — and the similarity map above it L16269 - **§176f** — THE DECLARATION FORM IS A MATCHING LEVER, SO RECONCILE TOWARD THE FORM THE MATCH NEEDS (P31 S52) L16967 - **§183** — THE DECLARATION-RECONCILIATION PLAYBOOK (P31 S53, measured on 20 byte-verified drafts) L18006 - **§193-H** — A pointer-derived base load `*(s32*)(p+K)` is uncacheable across ANY memory write or call (cse's third kill disjunct) — so the target emits one `lw` per store-separated RUN of spellings, and a run of stores sharing one `lw` is a C local you must declare L18852 - **§193-I** — A DECLARED AGGREGATE LOCAL HAS AN 8-BYTE FRAME STRIDE — CEIL(size,8), NOT size. N ARRAYS THEREFORE COST 8·k MORE THAN ONE HAND-LAID STRUCT (k = how many of them have size mod 8 ≠ 0), AND THE TELL IS IDENTICAL INSTRUCTION COUNT WITH EVERY $sp DISPLACEMENT SHIFTED BY THE SAME CONSTANT. L18892 - **§194-B** — A `addu $rA,$rB,$zero` copy feeding ≥2 `sh` stores is a SECOND, 16-BIT-DECLARED local — the width, not the clamp or the join liveness, is what keeps the copy (and func_80183094's pin + `"memory"` clobber are both provably inert) L19061 - **§194-D** — Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication + attribution correction of §165-17, NOT a new argument-register law L19162 - **§195-A** — §167-08's "an $aN READ before the jal is scratch" has a byte-proven FALSE-NEGATIVE class: an argument that DIES at the call is allocated straight into $aN, so its only def is a plain load far above the jal and every intervening use reads $aN — there is no positive tell in either direction, only the two-arity A/B L19768 - **§195-C** — A call-argument `%hi/%lo` pair sitting at the block head, far above its `jal`, is a load-delay-gap filler chosen by SOURCE STATEMENT ORDER — swap the two independent statements nearest the call; a single-use `void *p = &SYM;` call-arg temp is OUTPUT-inert against it (but NOT expand-stream-inert) L19881 - **§195-D** — §195 — `masked_diff.mask_for` returns 0 for EVERY `j`/`jal` word, so an internal `j` destination is invisible to match_one, the permuter scorer AND every similarity tier: a control-flow semantic error (which calls execute) surfaces as a 1-instruction residual mislabelled `DELAY-SLOT / profile=schedule`, or as an outright false MATCH L19929 - **§195-F** — fold-const.c:4825 canonicalises a `?:` whose THEN arm is zero (or constant against a non-constant ELSE) by inverting the condition and swapping the arms — so a ternary's written arm order is byte-inert there, and `c ? 0 : X` is unspellable: it always compiles as `!c ? X : 0` L20055 - **§195-G** — §NEW — TWO ARMS CALLING THE SAME CALLEE MERGE INTO ONE `jal` UNLESS EACH ARM'S OWN CODE CONSUMES THE RESULT: the C dial that keeps two call sites apart is WHERE the consumer test lives, and the branch SENSE you spell it with is byte-inert (jump.c:1737 canonicalises it) L20107 - **§195-H** — §165-27g — ON A FIXED-SYMBOL GLOBAL, `extern T D[];` + `D[0]` IS THE CSE RELOAD DIAL, AT ZERO ADDRESSING COST, AND THE ELEMENT COUNT IS INERT (§165-27's `T v[2]` CAVEAT IS A LOCAL-FRAME FACT AND DOES NOT TRANSFER) L20155 - **§196** — PUT ON THE CARD WHAT THE TREE ALREADY KNOWS: the fleet's declaration consensus (P31 S54) L20557 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§199-G** — At TWO case nodes the switch-vs-if oracle is not blind — but the tell is ALL tests positive + a trailing `j default`, NOT the first test's polarity L21084 - **§200** — THE ALIAS IS THE UNIVERSAL DECLARATION ESCAPE: stop negotiating with the TU's spelling (P31 S55) L21150 - **§201-A** — §150-B applies to `decl_prior`'s DEF row: for an overlay-window symbol the banked "definition" is usually another overlay's function, and the card ranks it ABOVE the destination TU L21209 - **§201-D** — THE SIGNEDNESS OF A div/mod MAGIC IS DECIDED BY THE STATIC TYPE OF THE DIVIDEND TREE AFTER `get_narrower` STRIPS WIDENING CONVERSIONS — NEVER BY A PROVABLE RANGE. AN `& 0xFF` IS NOT A CONVERSION, SO IT NEVER FLIPS THE MAGIC; A DECLARED-UNSIGNED LOCAL *OR* A NARROWING CAST WRITTEN AT THE DIVIDE BOTH DO. L21372 - **§202** — THE ALIAS CARRIES A DEFINITION, NOT JUST A DECLARATION: the DEF-SIDE-RETURN wall (P31 S56) L21500 - **§204-C** — WHEN A LOCAL BUFFER'S ADDRESS IS PASSED TO A CALL, ITS SIZE IS A FACT ABOUT THE CALLEE'S BODY, NOT ABOUT THE CALL SITE: grep the callee's proven definition and count the stores through the pointer parameter before you declare the local L21756 - **§204-E** — `decl_prior`'s `%hi/%lo` ARM HAS NEVER FIRED: the card's promised GLOBAL-TYPE row is 0 of 1,210 L21889 - **§208** — TWO NAMED LOCALS FOR ONE RELOADED EXPRESSION BUY TWO ALLOCNOS — the naming granularity owns the REGISTER SPLIT, not just the load count (P31 S58) L22253 - **§214** — THE BANKED TWIN MAY BE A MACRO, A DELETED `.s`, OR A SEMANTIC INVERSE — six ways a ≥0.9 similarity lies (P31 S58) L22565 - **§216** — DISTINCT ADJACENT SCALARS vs ONE ARRAY: one `lui` per access is the tell, and the array decl is UNUSABLE (P31 S58) L22667 - **§220** — THE PARAMETER ITSELF IS A REGALLOC DIAL: use it directly, prefer `s32` to `void*`, and place the save-copy AFTER the first call (P31 S58) L22789 - **§224** — CROSS-JUMP: WRITE THE DUPLICATE, AND READ A SHARED DELAY SLOT AS THE MERGE SIGNATURE (P31 S58) L22949 - **§236** — THE DECLARATION LAYER IS THE DOMINANT BANK-BLOCKER: NINE WAYS A BYTE-PERFECT BODY FAILS THE GATE (P31 S58b) L23437 - **§243** — SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P31 S58b) L23731 - **§250** — `%hi/%lo` vs `lw`: THE EXTERN'S ARRAY-vs-SCALAR SHAPE DECIDES ADDRESS MATERIALISATION (P31 S58b) L23983 - **§214** — addendum (P31 S58b) — FOUR MORE WAYS A HIGH-SIMILARITY TWIN LIES L24398 - **§226** — addendum (P31 S58b) — THE FRAME CATALOGUE: SEVEN MORE LEVERS, AND SLOT ORDER IS DECLARATION ORDER L24627 - **§263** — A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCHEDULE (P31 S59, byte-proven) L24937 - **ADD-10** — → §237 addendum (arity-evidence paragraph) — AN `la` PAIR ABOVE THE PROLOGUE `sw $ra` IS AN OUTGOING-ARGUMENT MATERIALISATION L25235 - **ADD-2** — → §236 addendum (item 1 corollary) — THE SILENT-SPLICE DECLARATION LADDER: WHERE THE TU HAS NO SPELLING, UNPROTOTYPED `()` BEATS THE FLEET VOTE L25377 - **§272** — The `(void)`-decl + empty-call wall: K&R the DEFINITION, not just the decls (P31 S59) L25609 - **ADDENDUM** — to §20 — a global declared as `T *` may itself BE the array base, not a pointer to dereference L25830 - **ADDENDUM** — to §172b-1 — TWO PLACEMENT DIALS THE PROMOTION LAW DOESN'T NAME: NARROW THE COUNTER'S DECLARED TYPE, AND MOVE THE CAST INTO THE LOOP BODY L26110 - **§280** — THE CURSOR'S DECLARED TYPE PICKS `sltu` vs `slt`: a `T*` bound test is UNSIGNED by C rule, an integer cursor is signed (P31 S60; wave cf, func_800CAE74, md_MAIN_031, byte-proven) L26966 - **ADDENDUM** — to §202 — the DEF-SIDE ALIAS also resolves a function-vs-DATA-symbol identifier clash, not only a function-vs-function prototype clash L27213 - **ADDENDUM** — §NNN — sharpens §37/§124 (unspecified-parameter-list family): DERIVE A CALLEE'S ARITY LOWER BOUND FROM ITS OWN `.s` STACK-ARGUMENT READS, NOT FROM ANY SINGLE CALL SITE L28336 - **§288** — A REGISTER PIN DECLARED UNINITIALIZED AND ASSIGNED ONLY AT ITS LATE, SOLE USE STILL FORCES THE FIXED REGISTER'S SAVE/RESTORE, AND THE ASSIGNMENT'S SOURCE POSITION CONTROLS WHERE THE VALUE MATERIALIZES (P31; `func_80186530`, ov_SC02_017, byte-proven, match_one MATCH re-verified) (P31 S60; waves #, byte-proven) L29100 - **§292** — DECLARING A SYMBOL UPSTREAM OF AN ALREADY-BANKED SIBLING THAT RELIES ON THAT SYMBOL'S IMPLICIT (K&R) DECLARATION CAN SILENTLY REPROTOTYPE THE SIBLING'S OWN CALL SITE (P31 S60; waves #, byte-proven) L29116 - **Addendum** — Counter zero in the DECLARATION slot, empty for-init — the prologue SHIFT-DRIFT/+K face of (func_8018275C) L30735 - **§324** — THE ARITY-SELF WALL NEEDS *BOTH* HALVES: A K&R DEFINITION **AND** A NO-PROTO TU DECL; EITHER ALONE STILL FAILS (P31 S67; rtu-proven ov_SC01_005/func_8017FBCC, 14 instances measured) L30959 - **§333** — FRAME SIZE IS SET BY *DECLARED* AGGREGATES, NOT USED ONES: AN UNREFERENCED TRAILING LOCAL IS A REAL DIAL (P31 S67; three independent byte-proven instances in one wave) L31125 - **§335** — AN `extern u16 A[]` READ AT A VARIABLE SUBSCRIPT ALLOCATES DEAD STACK TEMPS (~8B PER ACCESS) THAT INFLATE THE FRAME WHILE EMITTING ZERO EXTRA INSTRUCTIONS (P31 S67; byte-proven ov_SC06_025/func_8017EA74, closeness 141 -> 20) L31150 - **§342** — A `void *` PARAMETER CAST TO ITS REAL TYPE IN A LOCAL IS **NOT** BYTE-NEUTRAL WHEN A LATER PARAMETER ALSO NEEDS A CALLEE-SAVED REGISTER (P31 S67; byte-proven ov_SC07_002/func_80181394, NEW LAW) L31267 - **§343** — `decl_prior`'s FLEET MAJORITY CAN BE WRONG ABOUT THE TRUE SIGNATURE — READ THE RIVALS, NOT JUST THE WINNER (P31 S67; measured on func_8012BD14 / func_8012D624 / func_80143C74) L31284 - **§347** — LOOP REGISTER ASSIGNMENT IS A **DECLARATION-ORDER + LIVE-RANGE** DIAL: FIVE COMPOSABLE LEVERS, 178 -> 0 (P31 S67; byte-proven ov_SC06_029/func_8017EF34, 243 ins) L31352 - **§343-addendum** — SECOND INSTANCE OF THE WRONG-MAJORITY DECL (same function) L31400 - **§354** — THE giv **WORTH-WHILE TEST** IS A DIAL: RE-ASSOCIATE THE ADDEND INTO THE INDEX TERM AND `strength_reduce` DECLINES (P31 S68; byte-proven ov_SC03_105/func_801824CC, 320 ins, 201 → 5 → 0) L31552 - **§361** — ★ — A LOOP-TAIL BYTE SIGNATURE THAT NAMES ITS SOURCE SHAPE — AND THE LAW THAT A "SCHEDULING TIE" MAY BE YOUR OWN EARLIER LEVER (P31 S68; byte-proven main/func_800241C0, fable escalation, 19 → 0 in 3 iterations) L31647 - **§367** — RECONCILING A DECL CONFLICT BETWEEN TWO DRAFTS FOR THE SAME TU (P31 S68; main) L31787 - **§378b** — ★★★ — THE FOUR VARIANTS OF THE DECL BLOCKER, AND THE TWO PLACES §378 DOES **NOT** APPLY (P31 S69, all four measured the same day) L32582 - **§378c** — ★★ — THE FIFTH DECL-BLOCKER VARIANT: THE DRAFT REDECLARES WHAT THE TU ALREADY OWNS (P31 S69) L32732 - **§398** — ★★★ — `family_remap` CARRIES THE **SOURCE** TU's DECL ENVIRONMENT INTO A DESTINATION THAT ALREADY OWNS THOSE NAMES (P31 S69; measured 3 banked of 22) L32845 - **§415** — ★★ — A FILE-SCOPE DECL MAKES gcc-2.7.2 MERGE THE TU'S LATER *BLOCK-SCOPE* EXTERNS INTO IT (P31 S71; byte-proven `ov_SC04_011/func_80180B24`, 215 ins) L33557 - **§437** — ★★★ — A WRAPPED TRAILING COMMENT BROKE THE TU PARSER, AND THE §8b DECL LAYER WROTE `extern #define` (P31 S74; five independently-MATCHed jr bodies unblocked, `ov_SC06_029` byte-identical) L34431 - **§447** — ★★★ — THE BIGGEST FUNCTION LEFT IS A CARVE, AND THE VERDICT TOOL SAID "DECLARATIONS" (P31 S75; SaveLoadRoutine, 1,165 ins) L34864 - **§467** — GLOBAL-ALLOC TIES BREAK ON *DECLARATION* ORDER, AND A COPIED CLOBBER LIST IS A DEFECT L35366 - **§468** — THE `%lo`-FOLD EXTENDS TO STORES ONLY VIA `extern Struct SYM[]`, AND MASKING HID THE OPERAND ORDER L35390 - **§477** — ★★★ — THE `self_decl_tu` CLASS IS A SOLVED, MECHANICAL LANE: 16 DRAFTS, 16 BANKS (P31 S77) L35723 - **§481** — ★★★ — `conflicting types` IS A SAME-SCOPE ERROR; ACROSS SCOPES IT IS ONLY A WARNING (P31 S77, `main:func_8001FC08`, 400 ins) L35911 - **§487** — ★★★ — THE PSX LOADER'S PER-VERSION SIGNATURE SETS ARE A FREE PROVENANCE ORACLE: main's "WALL" BAND IS LIBPAD 4.2.1 (P31 S78) L36160 - **§495** — ★★★ — THE VERBATIM END-STATE (P31 S80 #10): TWO DEF-SIDE DECLARATION WALLS, A "BANK" THAT WAS THE ASSEMBLY, AND A GATE THAT DROPPED A BANK ON EXIT 0 L36630 ### jump tables & switches (72) - **§8** — rodata island (compiler jump tables) — the `.data→.rodata→.data` sandwich (Phase 7) L339 - **§8a** — rodata island in a flat OVERLAY — the tail sandwich, per matched jr-function (Phase 26 — PoC PROVEN) L361 - **§8a-pad** — a trailing `.word 0x00000000` under a jtbl dlabel is `.align` PAD, not an entry (Phase 26 session 6, byte-proven) L399 - **§8b** — MULTI-jtbl per overlay — the `ld_interleave --order` sandwich + the same-subseg cases (Phase 26 session 4) L423 - **§8e** — The jtbl ALIGNMENT LAW + the pad-spec filter — multi-table .rodata spans (Phase 29, byte-proven; `.run/probe_jtbl/verdict.md`) L549 - **§3-The** — clean object-level metric (use this, not the permuter score, for jtbl/rodata functions) L863 - **§41b** — T7 execution: the object-only probe OVER-counts BANKABLE by two link/rodata classes (Phase 25 T7-M1, 2026-07-10) L2847 - **§48** — The 12-core jr crack wave: the ALLOCNO-PRICING dials and the EBB rule (Phase 26 session 8, Ultracode, 9/12 MATCH first pass) L3443 - **§53** — SWEEP A FAMILY WITH THE TOOL ITS EXEMPLAR NEEDED: the jr/switch carve, and how omitting it manufactured the "families don't template" doctrine (Phase 28 T1, 2026-07-15) L4066 - **§61b** — The jtbl gate stage: built, and the ORDERING law it exposed (Phase 29 Task-14 stage 4, 2026-07-21) L4696 - **§61c** — The jtbl bank is INCREMENTALLY valid and CLEAN-INVALID (Phase 29, 2026-07-21) — the blocking finding L4751 - **§62** — The jtbl RECONCILE must also follow the carve: the post-carve draft reconcile (`harvest_verify._jtbl_reconcile`, Phase 29 SESSION-11, 2026-07-22) L4871 - **§81** — Banking a jr (jump-table) function: the 3-step carve chain, and why `match_one` cannot see the problem (Phase 29 SESSION-19, `func_8017C954`) L6361 - **§88d** — BANKING ORDER: run the §81 carve chain BEFORE banking, never after L6739 - **§97** — The gate's own tree hygiene: a refused carve, an unchecked recovery, and a snapshot that captured a dirty tree (Phase 29 SESSION-22) L7104 - **§105** — A gate's revert must survive an EXCEPTION, not just a failure (Phase 29 T53, `jtbl_family_bank`) L7469 - **§125** — Split the CARVE from the BODY before calling a jr residue a wall — and measure it by SHA from a CLEAN tree (P30 SESSION-28; **this section's first draft was WRONG and the method caught it**) L8244 - **§126** — The carve-within-a-carve: an ADDRESS RANGE is not an OPTIMIZATION REGION (P30 T2, byte-proven end-to-end) L8302 - **§129** — Post-carve, `rtu_match`/`match_one` COUNT THE JUMP TABLE AS INSTRUCTIONS; and a carve must never be committed without its owner (P30 S28, `func_8013BD74`) L8475 - **§129a** — the target instruction count is INFLATED after a carve L8479 - **§129b** — never commit a carve whose owner is still a stub (it strands the carve) L8500 - **§130** — An INCREMENTAL build can report BYTE-IDENTICAL for a change the CLEAN build cannot even LINK (P30 S28, the jr pair) L8524 - **§131** — The jtbl OVER-SPAN: `sltiu N` is ground truth in BOTH directions, and the zero-word rule only guards one (P30 S28, `func_80191C50`) L8571 - **§132** — The `JR-PAIR-IN-ONE-O0-OBJECT` "wall" was TWO instrument defects: a merged-double span the carve could not see, and a truncated object no rule deleted (P30 S29, `func_8013B83C` + `func_8013BD74`) L8608 - **§132b** — When the span's already-matched owner is ITSELF multi-switch: `--span-rel` (P30 S1, `func_8014032C`) L8698 - **§139** — A GATE THAT GREPS FOR VERDICTS MUST ASSERT 1:1 ACCOUNTING; and a `--src` filter must not survive a carve (P30 S38, wave 6: 10 of 16 drafts vanished) L9632 - **§161** — THE RETRY-WAVE HARVEST: a jump table indexed from zero, and two allocator traps (P30 S47) L11007 - **§179-F** — PINNING A LOOP-WALKED POINTER IS A TOTAL OFF-SWITCH FOR STRENGTH REDUCTION L17577 - **§188** — 🔴 THE `jr $ra` + `addiu $sp` TAIL IS AN **ASSEMBLER** ARTIFACT, NOT A FRAME SHAPE L18221 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§199-G** — At TWO case nodes the switch-vs-if oracle is not blind — but the tell is ALL tests positive + a trailing `j default`, NOT the first test's polarity L21084 - **§206** — THE JTBL-CARVE DRAFTING IDIOM: the bounds check is the entry count, and an EMPTY case owns a slot (P31 S56) L22163 - **§222** — SWITCH vs IF-CHAIN, PART 3: source arm order IS emission order, a leading EMPTY case buys the median split, and a 2-way dispatch with a shared post-block is a `switch` (P31 S58) L22845 - **§232** — WHEN THE `jr` DELAY SLOT'S STORE STORES THE RETURN VALUE, TIE THEM WITH ONE PSEUDO **(single observation — not yet cross-confirmed)** (P31 S58) L23231 - **§256** — GOTOS IN THE TARGET'S BLOCK ORDER REPRODUCE SWITCH PLACEMENT WITHOUT SWITCH'S SIDE EFFECTS (P31 S58b) L24124 - **§222** — addendum (P31 S58b) — IF-CHAIN vs SWITCH: THREE MORE DISCRIMINATORS L24528 - **§260-A** — STAGE 2 IS PROVEN, AND THE WHOLE jtbl PIPELINE IS AUTOMATED AT THE GATE (P31 S59, same day) L24823 - **ADDENDUM** — to §8 — WHEN `INCLUDE_RODATA` NEEDS A STANDALONE `.s` YOU CAN'T CREATE, CARRY THE FRAGMENT AS A FILE-SCOPE `__asm__` BLOB L26232 - **ADDENDUM** — §NNN — sharpens §55a / §164-37 / §165-28 (switch-vs-tree cluster): A SHAPE THAT LOOKS LIKE A SWITCH DISPATCH CAN BE PLAIN NESTED `if`s WHOSE SHARED BODY WAS TRIPLICATED BY THE SOURCE AND THEN CROSS-JUMP-MERGED BACK DOWN L28233 - **ADDENDUM** — §NNN — sharpens §162a3/§60b: A `sll $v0,16 / sltiu $v0,1` ZERO-TEST OF A JUST-DECREMENTED HALFWORD IS A SIGNED-TEMP WIDTH TELL OUTSIDE ANY SWITCH/RANGE-TEST CONTEXT L28303 - **§295** — THE KERNEL-TRAP STUB: §81's jr-DETECTOR WITHOUT A TABLE IS A PsyQ SYSCALL TRAMPOLINE — ROUTE TO §265, NEVER INTO THE CARVE CHAIN (P31 S61; wave m0a, 9 cards byte-proven; resolves §182's held cluster) L29411 - **§298** — THE SHARED-TAIL POSITION DIAGNOSTIC: A FOLDED TAIL *BETWEEN* SWITCH ARMS MEANS PER-ARM DUPLICATED STATEMENTS, NOT A HOISTED POST-SWITCH STATEMENT (P31 S61; waves g0e/g0f, `func_80181B68`, ov_SC06_016, byte-proven 68/68) L29552 - **§302** — A RED BINARY IS A DRIFTED SPEC, NOT A MYSTERY: THE THREE CARVE-STATE INVARIANTS AND HOW TO DERIVE EACH FROM THE BYTES (P31 S62 T2; five reds healed in one session, 5/5, +39 held banks) L29729 - **§304** — SELF-DEFINING RODATA: WHEN A FUNCTION'S `.s` IS THE ONLY OWNER OF THE DATA IT REFERENCES, THE C BODY MUST DEFINE IT (P31 S62 T3; byte-proven md_MAIN_011/func_800D04F4) L29809 - **§305** — "CARVE-REFUSED" AT GATE TIME IS THREE NAMED, DETERMINISTIC CLASSES — NONE OF THEM A CARVE (P31 S62 T3; 28 resolver drafts autopsied 28/28) L29832 - **§308a** — A TWO-INSTRUCTION ARM STUB LAID *BEFORE* THE OTHER ARM AND REACHED ONLY BY A TRAILING `j` SURVIVES ONLY BEHIND A SHORT-CIRCUIT `&&` GUARD PLUS A SEPARATE RE-TEST; EVERY PLAIN if/else-if, GOTO-LADDER, NESTED-INVERTED-if AND 1-/2-NODE switch LETS jump1 INLINE IT (−2 ins) (P31 S63 t5e-t5i; byte-proven func_80180808) L30022 - **REFUTED** — claim — REFUTED: a block shared across TWO switch statements is ordinary forward cross_jump, not a (func_8001B0D4) L30648 - **§NNN** — REFUTED: A CODE BLOCK SHARED BY TWO *DIFFERENT* SWITCH STATEMENTS IS ORDINARY FORWARD `cross_jump`, NOT A SHAPE C CANNOT REACH — §164-11 ALREADY BANKS IT FROM PLAIN NESTED SWITCHES (P31 S66 round3, `func_8001B0D4`; ⚠ **UNPROVEN** — gate-REFUSED draft, closeness 68→45, never MATCHed) L30650 - **§322** — A PROBE THAT ANSWERS A *NECESSARY BUT NOT SUFFICIENT* QUESTION WILL PRICE BLOCKED WORK AS FREE: RUN THE REAL PLANNER WHEN THE PLANNER IS PURE (P31 S67; measured, 96 of 159 open jtbl functions) L30876 - **§338** — `jtbl_carve._sltiu_bounds` MISREADS A NON-SWITCH `sltiu` AS A BOUNDS CHECK, OVER-SPANNING THE TABLE (P31 S67; ov_SC06_022/func_80185B80, byte-diagnosed) L31181 - **§339** — A 2-CASE SWITCH OMITS THE LOW-BOUND RANGE TEST, SO THE PRESENCE OF `slti/bnez` BETWEEN THE `beq`s IS A **COUNT TELL** FOR A THIRD CASE NODE (P31 S67; byte-proven ov_SC02_005/func_80190538, 197 ins) L31191 - **§362** — TWO TRAPS WHEN A CARVE MOVES A STUB INTO THE `-O0` TU (P31 S68; byte-proven, 6 fns / 2,547 ins across ov_MAIN_012 / ov_SC02_037 / ov_SC03_107) L31692 - **§371** — ★★ — CARVING A **SINGLE-OBJECT MODULE BINARY**, AND THE spimdisasm RODATA-MIGRATION TRAP THAT FOLLOWS (P31 S68; byte-proven md_MAIN_003, func_800D0D6C 345 ins) L31900 - **§384** — ★★★ — A CARVE-CONFIG BANK IS RED UNTIL YOU RE-EXTRACT, AND THAT LOOKS EXACTLY LIKE A FALSE BANK (P31 S69; measured twice, cost one destroyed match) L32269 - **§387** — ★★ — **SPLIT-FOLD DISPATCH CLOBBER**: one switch case needs a reload, another must keep the fold (P31 S69; byte-proven main/func_80030F80, 343 ins, escalation 3 → 0) L32371 - **§322b** — ★★★ — THE CARVE CLASS IS COMPLETABLE, AND EVERY WORKTREE `CARVE-REFUSED` WAS AN INSTRUMENT VERDICT (P31 S69, Fable-3 audit; byte-proven end-to-end) L32677 - **§3-A.** — THE ORACLE HAS A HOLE: match_one cannot see a jump table (§405-A) L33108 - **§412** — ★★★ — §323 CARVE BLOCKER 2 WAS A REGEX THAT COULD NOT SEE PAST `__attribute__` (P31 S71) L33434 - **§426** — ★★★ — main's SWITCH FUNCTIONS WERE NEVER A CODEGEN WALL: ONE RODATA CARVE HAD BEEN MISSING SINCE PHASE 7 (P31 S72; 3 of the 11 "PROVEN gate-rejects" banked byte-identical in 14 s) L33865 - **§428** — ★★★ — A ZERO-BYTE CROSS-JUMP BARRIER: ADVANCE THE POINTER INSIDE EACH SWITCH ARM (P31 S72; `main/func_80026D64`, 218 ins, MATCH in 2 compiles) L33964 - **§431** — ★★★ — SPLITTING A 27,000-LINE TU AT ITS ORIGINAL BOUNDARIES: THE JTBL SPANS TELL YOU WHERE, AND THE COMPILER TELLS YOU WHAT CROSSES (P31 S72; `src/800.c` -> `800.c`/`800_b.c`/`800_c.c`, byte-identical with nothing banked) L34132 - **§433** — ★★★ — ON A SWITCH FUNCTION, CASE SOURCE ORDER IS THE DOMINANT RESIDUAL — AND `match_one` IS BLIND TO IT (P31 S73; 4 of 5 consecutive main MATCHes) L34195 - **§435** — ★★★ — AN OVERLAY TU SPLIT IS NEAR-FREE, AND THE GAP TEST IS "IS THIS WORD A CODE ADDRESS", NOT "IS IT ZERO" (P31 S74; four overlays split, all byte-identical, CARVE-BLOCKED class emptied fleet-wide) L34317 - **§437** — ★★★ — A WRAPPED TRAILING COMMENT BROKE THE TU PARSER, AND THE §8b DECL LAYER WROTE `extern #define` (P31 S74; five independently-MATCHed jr bodies unblocked, `ov_SC06_029` byte-identical) L34431 - **§440** — ★★★ — A `.rodata` CARVE PIECE BINDS TO A SUBSEG, NOT TO A FUNCTION: EXTEND THE CARVE INSTEAD OF ISOLATING (P31 S74; 6 banked, resident included) L34586 - **§444** — ★★★ — A REJECTED GATE DAMAGED COMMITTED CARVE STATE, AND `git status src/` SAID NOTHING WAS WRONG (P31 S75) L34739 - **§446** — ★★★ — "RELOC-ONLY REMAP GATED DIFF" IS A CARVE VERDICT, NOT A CODEGEN VERDICT, UNTIL YOU DIFF THE `.text` (P31 S75; 4 siblings + 1 latent, ~1,300 ins) L34817 - **§447** — ★★★ — THE BIGGEST FUNCTION LEFT IS A CARVE, AND THE VERDICT TOOL SAID "DECLARATIONS" (P31 S75; SaveLoadRoutine, 1,165 ins) L34864 - **§3-B.** — A SINGLE-SET LOCAL'S VALUE IS VISIBLE AT A SWITCH JOIN, AND THAT ERASES A ZERO-EXTENSION (CONFIRMED: `ov_SC03_105:func_801806F8`, 241 ins, banked from closeness 235) L34985 - **§3-D.** — AN OFFLINE JTBL-RODATA PLACEMENT AUDIT (CLAIMED: `md_MAIN_034:func_800CB00C` — did NOT bank, and that is the point) L35018 - **§491** — ★★ — THE MECHANICAL LEFTOVERS (P31 S79 #6): A PHANTOM STUB, TWO JTBL TWINS, ONE EXACT CLONE — AND THREE TOOL GAPS THE BANKS EXPOSED L36356 - **§498** — ★★★ — A CARVE THAT DROPS THE BINARY'S `--pre` CLAUSE, AND A GATE THAT IGNORED THE EXTRACT'S EXIT CODE: HOW A BYTE-CORRECT DRAFT WAS BOOKED "DIFF" (P32 T1a; resident `func_800D128C` BANKED 243 ins after two instrument fixes) L36728 ### optimisation level (-O0/-O2) (26) - **§6** — Per-module optimization mixing — the -O0 boot module (Phase 7) L265 - **Detecting** — the opt level (do this first) L273 - **Build** — mechanism — per-file opt override (splat resegmentation) L307 - **§18** — Per-file `-O0` split inside an overlay/blob (Phase 19 T1) L1542 - **§38** — The WHALE `func_80144B9C` (770 ins): the -O0 struct-assign memcpy idiom + the -O0 reach-134 ×134 rollout (Phase 24 T7 §G, cheap Opus — no Fable5, no calls.c) L2550 - **§39** — The ×1→×134 giant-endgame: propagate a matched **-O2** giant via the NATIVE DEFINE-macro path (Phase 24 T7 §G close, 2026-07-08) L2560 - **§116** — Optimization level is a property of the FILE, not the function: read a family 0/N against the member's stub HOME (Phase 29 T79) L7915 - **§3-The** — finding: opt level is per FILE, so the file's contents must be opt-HOMOGENEOUS L8317 - **§127** — The `-O0` regime: the CONSTANT-OFFSET FOLD, and why `-O0` needs its own idiom set (P30 T3 wave, 15 targets) L8387 - **§3-The** — rest of the `-O0` regime (write PLAIN C, and mean it) L8405 - **§127a** — §71 (sibling-first) is the strongest `-O0` lever, and it beats the index L8415 - **§132** — The `JR-PAIR-IN-ONE-O0-OBJECT` "wall" was TWO instrument defects: a merged-double span the carve could not see, and a truncated object no rule deleted (P30 S29, `func_8013B83C` + `func_8013BD74`) L8608 - **§195-K** — At -O2 the §18 array-of-struct lever is a FRAME lever, not a length lever — but only when the N reads carry DISTINCT index expressions; with a SHARED index §18's +2-instruction residual is still alive at -O2 (the submitted "memory-loaded narrow index" precondition and unconditional length-neutrality are both falsified) L20311 - **§261** — THE -O0 ORACLE: DERIVE THE OPT LEVEL FROM THE TARGET, AND `$fp` IS NOT THE TELL (P31 S59) L24859 - **§261a** — THE -O0 FRAME-RELOAD GRAMMAR: reload COUNT disambiguates the C spelling (P31 S59, byte-proven) L24885 - **§265** — THE VERBATIM-ASM BANK LANE: A FUNCTION NO -O2 C CAN EVER MATCH BANKS AS A RAW `__asm__` BODY (P31 S59b; two banked cards, two in-tree precedents) L25028 - **ADDENDUM** — to §127 — the -O0 constant-offset fold keys on the MEMBER-ACCESS tree shape, not on the index being constant L29180 - **ADDENDUM** — to §261a — FOUR MORE -O0 DIALS BEYOND THE RELOAD COUNT (boot.c + md_MAIN_011/003, wave m0a/g0a/g0f, byte-proven) L29208 - **Addendum** — Addendum to §261a — the -O0 increment-operator copy also fires on the POINTER LOCAL itself (func_8001212C) L30483 - **Addendum** — Addendum to §164-XX/§165-06 — at -O0 a bare local IS a memory lvalue, so the increment ope (func_8001212C) L30487 - **§362** — TWO TRAPS WHEN A CARVE MOVES A STUB INTO THE `-O0` TU (P31 S68; byte-proven, 6 fns / 2,547 ins across ov_MAIN_012 / ov_SC02_037 / ov_SC03_107) L31692 - **§388** — ★★★ — THE **-O0 COLOURING ORACLE**: simulate `stupid.c` instead of grinding spellings (P31 S69; main/func_80011380 proved a C-level WALL at 6) L32394 - **§466** — `main` (509 ins, -O0): ADDRESS CONTEXT EMITS `mult` INDEX-FIRST L35320 - **§474** — A *PROVED* C-LEVEL FLOOR: `split_tree` + `stupid.c` (-O0), from `main:func_80011380` L35554 - **§486** — ★★★ — CARVING AN `-O0` ISLAND IN **main**: FIVE COUPLED PIECES, AND THE TWO THAT ANNOUNCE THEMSELVES (P31 S77) L36115 - **§492** — ★★ — "C-PLUMBING" WAS THREE DIFFERENT THINGS (P31 S79 #7): A RAW SPLICE THE GATE'S LADDER BROKE, TWO -O0 BODIES THE CHECKER COMPILED AT -O2, AND TWO DRAFTS THAT BELONGED TO OTHER OVERLAYS L36402 ### family propagation & sweeps (123) - **§8d** — Templating a body INTO a TU must not CHANGE its declaration environment — demote the carried data externs (Phase 26 session 8, byte-proven on `func_8015AE2C` ×133) L502 - **§11** — Cross-binary dedup & code-sharing (Phase 11 — "one match unlocks many") L927 - **§3-The** — mechanism: game-code dedup is SOURCE-LEVEL, not an object swap (R-D1, the key lesson) L945 - **§3-The** — flat-blob overlay config (what the template encodes) L1090 - **§3-THE** — NON-4-ALIGNED-OVERLAY GOTCHA (≈75% of the fleet; fixed in the template + Makefile, automatic) L1098 - **§14** — Propagate a matched function across the fleet — `tools/dedup_propagate.py` (Phase 15) L1145 - **§19** — Scaling the toolkit waves — the recovery PIPELINE + the propagation CAP (Phase 19 T3) L1620 - **§24** — The `ov_SC01_077_a.c` split-file vein: split-aware propagation, but loose-typing-gated (Phase 21) L2108 - **§25** — The "schedule" class is mostly COALESCING (pin-crackable), not scheduling; + the gate two-stage + h_exact over-counts ×134 (Phase 21, cont.6) L2155 - **§3-h** — _exact OVER-COUNTS ×134 — verify shareability before crediting a class's "reach-134" (R14) L2200 - **Where** — this leaves the reach-134 tail (cont.6 option-3, now CONFIRMED byte-backed) L2256 - **§28** — Banking a "close=0 gate-rejected" giant: the canonical-extern recovery (Phase 22 T1, byte-proven on func_8015126C ×134) L2348 - **§28c** — The close=0 recovery is NOT fully exhausted (§26 corrected, R14); + the dedup_propagate registry-skip recovery (Phase 22 T2) L2375 - **§28b** — The struct-walled close=0 giant (§28 case #3) IS bankable: the engine_types.h type-lift (Phase 22 T2, byte-proven on func_80156B74 ×134) L2380 - **§28d** — The "macro-extern-injection" lever: freeing reach-134 inline matches dedup_propagate skips as "not self-contained" (Phase 23, `tools/inject_capped_externs.py`) L2389 - **§30** — Fable5Max cracks §20/§10 "unsteerable" from the gcc SOURCE: store-vs-load is a `/s` aliasing flag, the def-side wall has a macro escape, + the birthing-boost (Phase 23, byte-proven on giant `func_8014EE14` 248 ins ×134) L2405 - **§34** — The `func_80138ED0` giant crack: gcc-2.7.2's **3-qty sort bug** + the **zero-byte asm allocation toolkit** + the **giv-init fence** (Phase 24 T5; Opus→close=21, Fable5→MATCH ×134) L2477 - **§38** — The WHALE `func_80144B9C` (770 ins): the -O0 struct-assign memcpy idiom + the -O0 reach-134 ×134 rollout (Phase 24 T7 §G, cheap Opus — no Fable5, no calls.c) L2550 - **§39** — The ×1→×134 giant-endgame: propagate a matched **-O2** giant via the NATIVE DEFINE-macro path (Phase 24 T7 §G close, 2026-07-08) L2560 - **§40** — Structural families: the MECHANICAL symbol-remap (crack one exemplar → remap the rest, ~0 tokens) (Phase 25 T3/T7, 2026-07-08) L2576 - **§40a** — The DECL-RECONCILE pass (type-lift so remapped drafts compile in the sibling TU) (Phase 25 T7.2, 2026-07-08) L2624 - **§40b** — The reloc-tracker blind spot: the indexed-global idiom that hid the "reach-1 tail" (Phase 26 Task 1, 2026-07-11, byte-verified V0/V1) L2657 - **§40c** — The h_seq per-sibling reconcile: templating a reconcile-class crack ×134 (Phase 26 Task 8, 2026-07-12, byte-proven) L2699 - **§41c** — T7-M2: the ×134 def-side-wall sweep via per-sibling RE-reconcile (Phase 25, 2026-07-10; 4,389 banks, ~0 agent tokens) L2877 - **§42** — The F-band ≤28 regalloc crack wave: register-pin/DENSITY levers beat the permuter (Phase 25 T7 F-band, 2026-07-10; Ultracode 9-worker wave, 4/9 banked byte-identical, 266 swept ×134) L2938 - **§42e** — propagating a CRACK ×134: the def-finder bug + the byte-drift residual (the "remap-fail" misdiagnosis) L3159 - **§43** — The K&R s16-param definition DISSOLVES the "narrow-param wall" for by-value register args (Phase 25 task A, Fable5 crack of the 369-ins giant `func_80166994` ×134, 2026-07-11) L3233 - **§45** — The flagship `func_80133CD4` crack (399 ins ×134): the merged-variable permutation-breaker + the 1-death local-alloc gate (Phase 25 task A giant escalation, Fable5 gdb-on-cc1, 2026-07-11) L3336 - **§46** — The `func_80178D40` crack (890 ins ×134, the heaviest core in the game): four LOOP-STRUCTURE levers cheap-Opus found by reading loop.c/jump.c/cse.c (Phase 26 session 8, 2026-07-13) L3351 - **§47** — The live-length SLIDER: splitting a global.c allocno-priority TIE with one zero-byte asm (Phase 26 session 8, Fable5 Max, byte-proven on `func_8017BEBC` 952 ins ×113) L3403 - **§49** — The LUID DIAL: a zero-byte SCHEDULING dial (the sched.c analogue of §47) — `func_8017A4AC` (536 ins ×134), Phase 26 session 8 L3565 - **§52** — The WALKER-FAMILY skeleton: 6 regalloc-order levers + a deeply-characterized intrinsic wall (`func_80178004`, 165 ins ×134; Phase 26, Fable5, 2026-07-15) L3951 - **§3-The** — 6 levers (the "walker-family skeleton" — apply to the 12 siblings; levers 1-5 retire ~half the residual) L3962 - **§53** — SWEEP A FAMILY WITH THE TOOL ITS EXEMPLAR NEEDED: the jr/switch carve, and how omitting it manufactured the "families don't template" doctrine (Phase 28 T1, 2026-07-15) L4066 - **§55c** — Sizing the propagate: a TARGETED propagate is ~4 min/core, and "it's slow" was a BROKEN-TREE ARTIFACT L4179 - **§54** — `--fix-def-sig`: the member's CANONICAL DECLARATION is a build step too (tiny-IMM mega-pools, +4,801, Phase 29 T6, 2026-07-16) L4201 - **§56** — Banking a hand-drafted GIANT into its exemplar TU: self-contained draft vs live-TU decls, 4 reconciliation tactics (`func_8013FAF8`, 312 ins ×138, Phase 29 T4, 2026-07-17) L4225 - **§56b** — PROPAGATING an h_seq giant: the exemplar's externs MUST be fleet-canonical, not the draft's types L4270 - **§57** — The SELF-decl normalize: the sibling's OWN caller declares the templated fn divergently (the third §17a-1 direction; `tools/normalize_self_decls.py`, func_801670E4 ×137, Phase 29, 2026-07-18) L4294 - **§57a** — Two NSD corrections + the SURGICAL-ONLY law + the honest broad-sweep yield (Phase 29, 2026-07-18) L4341 - **§59** — Three h_seq sweep-residual classes match_one/the-exemplar-bank don't reveal (Phase 29 crack-wave close, 2026-07-18) L4415 - **§3-Two** — corollaries worth remembering L4488 - **§65a** — The blast-radius taxonomy (makes §61's law structural instead of remembered) L5068 - **§3-Giv** — record order (the §70 family) L5798 - **§75** — A propagation cap is usually a MINORITY-SPELLING SOURCE OVERLAY, not a wall: census the carried extern before believing the exclusion message (Phase 29 SESSION-19, `func_8014F3E8` ×4 → ×138) L5904 - **§75a** — The exclusion classes, enumerated with named causes (Phase 29 SESSION-19, the 134-binary `dedup_extend` sweep) L5963 - **§75b** — A body's preamble can carry `#define`s, not just `extern`s; extraction lifts only the externs (Phase 29 SESSION-19, `func_80165CA0` ×3 → fleet) L5986 - **§76** — The allocno CLASS (local vs global) is the dominant regalloc lever, and C reaches it ONLY through declaration scope and variable reuse (Phase 29 SESSION-19, behemoth #3 `func_8017F510` 1,511 ins, 97 → MATCH, pin-free) L6066 - **§77** — Every extraction tool carries a NARROW hard-coded preamble set; anything outside it silently caps the body's reach. Diff the preamble before you gate. (Phase 29 SESSION-19 — three variants in one session, two different tools) L6122 - **§84** — The DERIVED-OFFSET remap bug: a hand-computed literal that encodes the DISTANCE between two per-overlay symbols, and why `match_one` is structurally blind to it (Phase 29 SESSION-20, `func_8013D53C`) L6510 - **§86** — Pinned-exemplar templatability is a PER-FAMILY property, not a per-member rate; and the §42e pin guard is now over-conservative (Phase 29 SESSION-20) L6627 - **§3-THE** — LAW: all-or-nothing PER FAMILY L6638 - **§3-Why** — the two live families differ from the three dead ones — the open question L6665 - **§3-The** — cheap discriminator, before spending a sweep L6704 - **§89b** — the parallel gate farm existed; the family path could not reach it (`tools/sweep_parallel.py`) L6783 - **§91** — A structure-TRANSFER is only valid where the structure corresponds: the `--like` role trap (Phase 29 SESSION-21, `func_8012AAAC` ×137) L6880 - **§94** — A family sweep's `0/N` is a TYPE-CARRY failure until proven otherwise: lift the exemplar's local types, ALL of them, transitively (Phase 29 SESSION-21, `func_8016B6BC` 0/137 → 137/137) L6985 - **§105** — A gate's revert must survive an EXCEPTION, not just a failure (Phase 29 T53, `jtbl_family_bank`) L7469 - **§107** — A lever wired into ONE gate path is a lever most families cannot reach (Phase 29 T56, `func_80144090` 0/136 → 136/136) L7551 - **§108** — Diagnosing a family `0/N`: the four causes, and the third opt-in lever (Phase 29 T59) L7589 - **§111** — The distinct-code metric is not noisy: a family pays it only if its members are byte-VARIANT (Phase 29 T66) L7719 - **§116** — Optimization level is a property of the FILE, not the function: read a family 0/N against the member's stub HOME (Phase 29 T79) L7915 - **§123** — PROPAGATE A FAMILY WITH THE TOOL ITS TIER NEEDS: `dedup_propagate` is h_exact-only; its refusals are statements about the TOOL (P30 wave 1, 2026-07-30) L8154 - **§124** — A "not matched" verdict can mean the definition is there under a DIFFERENT C NAME: the asm-label alias blind spot (P30 SESSION-28, `func_8016191C` ×137) L8191 - **§124a** — a family sweep's `0 matched-exemplar families` may be a FILTER, not a wall L8236 - **§134** — MULTI-LINE BLINDNESS: one root cause, four faces, in `family_remap`'s preamble scanner (P30 S6, 190 zero-crack families) L8749 - **§136c** — SIBLING-FIRST is a DERIVATION shortcut, not just a conflict fix (the fastest route in a family wave) L9107 - **§138** — The propagation lanes: a gate refusal is a DECLARATION, and which lever you owe depends on blast radius L9453 - **§142** — An open stub whose `h_exact` class is MATCHED elsewhere is FREE. Propagate the body; do not gate a draft. (P30 S39, +7,710 ins in two commands) L9816 - **§143** — `cast_call_sites` read a RETURN STATEMENT as a prototype and deleted it. A 0/39 sweep became 18/39. (P30 S40) L9872 - **Then** — propagation returned 0/137 TWICE — both times a missing TYPE L10038 - **§3-A.** — The frame has THREE strata, and stratum 3 is unreachable from C L10079 - **§3-E.** — A `qty_compare` TIE is not spelling-reachable — recognise it and stop L10121 - **Consequence** — for the family (a real scheduling decision) L10133 - **§150** — A register ROTATION across symmetric blocks is VARIABLE-IDENTITY evidence, not an allocator tie (P30 S43, `func_8017C6F4`, 947 ins ×4) L10350 - **§151** — THE GHOST WEDGE: when a load-before-store transposition is unreachable by ANY statement order (P30 S43, `func_8017EF68`, 969 ins) L10405 - **When** — to reach for it L10445 - **§152** — BYTE SIZE is the family key that name- and h_seq-grouping both miss (P30 S43, the 0xECC family: 1 crack → 12 overlays → 11,364 ins) L10456 - **§156** — an ORPHANED reconcile poisons the fleet: `dedup_propagate`'s kept edit (S45 p6/p7) L10678 - **§160** — THE REACH-15 WAVE HARVEST: an align-1 block move, and the instrument that called it a wall (P30 S47) L10931 - **§162** — S48 WAVE-1 HARVEST (P30, 2026-08-11): the reach-ordered sibling campaign's first 12 targets L11052 - **§163z** — THE UNVETTED REMAINDER (do not cite as law; each needs a dedupe pass) L11866 - **§3-The** — ADDRESS-CLASS TABLE: which load/store pairs even REACH the `/s` clause (P30 S48 wave 4, `func_80185B44`, ov_SC03_014) L14400 - **§168** — THE COUSIN TIER (P30 S49, 2026-08-12): h_seq's exact-hash brittleness, measured — and the similarity map above it L16269 - **§171a** — THE MECHANICAL A-PROP DRAFT (P30 S50): 256 members banked with no agent in the loop L16478 - **§171b** — THREE CARRIES THE MECHANICAL DRAFT NEEDS (P30 S50, banking the top-reach families) L16528 - **§3-G.** — TWO MODELLING TRAPS THAT COST THESE AGENTS SWEEPS OF HUNDREDS OF COMPILES L17355 - **§193-E** — A varying-address (pointer) load is re-emitted once per CSE-LIVE INTERVAL, and naming it in a C local is the only C-level lever over that count — no store SPELLING has any reach L18697 - **§194-E** — The wave card's `exemplar` is the TARGET ITSELF on 42/73 wave-U and 36/71 wave-T cards — a construction consequence of `atlas.py:657` (exemplar = max-nins OPEN member) meeting `build_wave_atlas.py:166` (one card per gid); and the shipped §193-A `seed_ref` is same-binary on 0/51, so no card field can ever name a destination-TU sibling L19203 - **§195-E** — A 0/1 materialised at a JOIN immediately before the controlling `beqz`/`bnez` proves the source NAMED the condition — a truth expression in an `if`'s controlling position reaches `do_jump`, which has no value path L19997 - **§199-B** — A permutation sweep that holds ANY statement fixed is not a sweep: the statement an agent pins as "obviously load-bearing" is the one carrying the signal, and the partial sweep returns a FLAT residual that reads as proof of order-invariance L20818 - **§203** — A DEDUPED TYPEDEF MUST PRECEDE EVERY SPLICE POINT, NOT JUST ITS OWN (P31 S56) L21537 - **§225** — THREE CONTROL-FLOW SHAPES NO STRUCTURED SPELLING REACHES (P31 S58) L22985 - **§30** — addendum (P31 S58b) — THE ANONYMOUS STRUCT MEMBER REF GRANTS `/s`, AND THAT IS A TWO-FOR-ONE L24209 - **§3-1a.** — The §266 sweep — every solo-lever A/B run for this batch L25536 - **ADDENDUM** — to §179-D (GTE macro reference family) — `gte_SetRotMatrix`/`gte_SetTransMatrix` bodies L25727 - **ADDENDUM** — to §82 — struct copies must stay MEMBER-WISE, not block-moved, to match a spill-slot save L25753 - **ADDENDUM** — to §136d-1 (RC-12, the `$0`-add / opaque-copy family) — two symptoms beyond "compare reads the wrong register" L25791 - **§282** — gcc's OWN LOOP REVERSAL PUTS THE COUNTER INIT AFTER THE HOISTED MOVABLES — a position no hand-written down-count can reach (P31 S60; wave cf, func_80180DD8, ov_SC04_005, byte-proven) L27046 - **ADDENDUM** — to the zero-byte-asm-slider family (§47 / §148-C / §153) (func_800CB874, md_MAIN_040 — wave dj; open tension with a more cautious dm-wave card — see closing) L27965 - **ADDENDUM** — the zero-emission-asm family gains a REF-SLIDER PLACEMENT LAW and a paired RESTORER (func_800CB900, md_MAIN_026) L28175 - **ADDENDUM** — §NNN — sharpens §37/§124 (unspecified-parameter-list family): DERIVE A CALLEE'S ARITY LOWER BOUND FROM ITS OWN `.s` STACK-ARGUMENT READS, NOT FROM ANY SINGLE CALL SITE L28336 - **ADDENDUM** — to §199-F family (func_8017EFB0, ov_SC02_021, wave cu) L28488 - **§287** — A STACK-FRAME HOLE BELOW TWO ADDRESS-TAKEN AGGREGATE LOCALS IS A LEADING PAD MEMBER OF ONE COMBINED STRUCT, NOT SEPARATE LOCALS OR A REGISTER PIN (P31 S60/dj; `func_8017D104`, ov_SC06_027, byte-proven 47/47) (P31 S60; waves #, byte-proven) L29096 - **ADDENDUM** — to §127 — the -O0 constant-offset fold keys on the MEMBER-ACCESS tree shape, not on the index being constant L29180 - **§296** — THE FRAME CHECK OUTRANKS THE ATLAS LEVER: READ PROLOGUE/EPILOGUE BEFORE DRAFTING ANY C — A REAL TELL CAN LIVE INSIDE AN UNREACHABLE FRAGMENT (P31 S61; wave m0a, 16 cards byte-proven; extends §179-C) L29466 - **§297** — ONE GIV SERVING MIXED-WIDTH LOADS *AND* A STORE NEEDS STRUCT-MEMBER SPELLING; CAST/INDEX SPELLINGS RE-SPLIT THE PSEUDO (P31 S61; wave m0a, `func_800347C8`, main, byte-proven 31/31) L29509 - **§307** — THE BRUTE-FORCE-THE-STATEMENT-ORDERS LEVER HAS A BOUND: A FAN-OUT COPY'S PRIORITY IS SCHEDULER-INTERNAL, AND NO SOURCE ORDER REACHES IT (P31 S63; byte-evidenced NEGATIVE result, main wave) L29949 - **§308a** — A TWO-INSTRUCTION ARM STUB LAID *BEFORE* THE OTHER ARM AND REACHED ONLY BY A TRAILING `j` SURVIVES ONLY BEHIND A SHORT-CIRCUIT `&&` GUARD PLUS A SEPARATE RE-TEST; EVERY PLAIN if/else-if, GOTO-LADDER, NESTED-INVERTED-if AND 1-/2-NODE switch LETS jump1 INLINE IT (−2 ins) (P31 S63 t5e-t5i; byte-proven func_80180808) L30022 - **§314** — AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES ALL COMPILE BYTE-IDENTICAL, AND NONE REACHES A TARGET THAT SINKS THE NEGATIVE ARM INTO AN OUT-OF-LINE BLOCK (P31 S65; UNPROVEN, gate-refused draft func_80181720) L30251 - **§NNN** — REFUTED: A CODE BLOCK SHARED BY TWO *DIFFERENT* SWITCH STATEMENTS IS ORDINARY FORWARD `cross_jump`, NOT A SHAPE C CANNOT REACH — §164-11 ALREADY BANKS IT FROM PLAIN NESTED SWITCHES (P31 S66 round3, `func_8001B0D4`; ⚠ **UNPROVEN** — gate-REFUSED draft, closeness 68→45, never MATCHed) L30650 - **§323** — CARRYING FILE-SCOPE TYPES BETWEEN SPLIT TUs: DEDUPE BY NAME, SKIP HEADER-PROVIDED, AND THE GUARD TRAVELS WITH THE BLOCK (P31 S67; ov_SC02_000, two defects fixed, one open) L30906 - **§341** — AN HImode STORE TEMP REWEIGHTS A sched2 TIE-BREAK THAT NO STATEMENT ORDER CAN REACH (P31 S67; byte-proven ov_SC03_006/func_801823B8, last 4 ins) L31253 - **§355** — A REMAPPED SIBLING'S **SOURCE BIAS IS NOT ITS EMITTED BIAS** — DO NOT HAND-SHIFT OFFSETS TO MATCH THE ASM (P31 S68; byte-proven ov_SC07_007/func_8013DD68, 187/187 in 2 iterations) L31570 - **§368** — ★★★ — THE **RELOAD-REMAT CONSTANT**: REACH A REGISTER NO PIN CAN REACH (P31 S68; ov_SC03_105/func_80187A30, 339 ins, fable escalation closed 8 → 0 in ONE edit) L31802 - **§379** — ★★★ — **MEM_IN_STRUCT_P**: THE SAME LOAD, WRITTEN AS A STRUCT MEMBER, SCHEDULES WHERE A CAST CANNOT (P31 S69; byte-proven main/func_80021284 220 ins and main/func_8002D904 217 ins, found INDEPENDENTLY by two agents) L32174 - **§396** — ★★★ — SIX LEVERS FROM THE S69 SINGLETON ROUND, INCLUDING ONE THAT ONLY A `COND_EXPR` REACHES (P31 S69) L32747 - **§398** — ★★★ — `family_remap` CARRIES THE **SOURCE** TU's DECL ENVIRONMENT INTO A DESTINATION THAT ALREADY OWNS THOSE NAMES (P31 S69; measured 3 banked of 22) L32845 - **§3-B.** — THE SCHEDULER DIALS (the dominant residual family this wave) L33118 - **§408** — ★★★ — §406 REFUTED AS A SWEEP: THE SHAPE IS THE FAMILY, THE DISAGREEMENT IS THE DEFECT (P31 S71; 0 MATCH / 14 applied, 0 / 210) L33219 - **§420** — ★★★ — A MULTI-CLUSTER SYMBOL REBASE, AND THE BARE-NAME DEDUP THAT HID THREE QUARTERS OF IT (P31 S71; 4 banked in 57 s) L33702 - **§421** — ★★★ — A `la $tN` + `addiu` PAIR CAN BE A **RELOAD** ARTIFACT THAT NO C SPELLING REACHES (P31 S71; byte-proven `md_SC07_003/func_801A293C`, 313 ins, 6 → 0) L33739 - **§446** — ★★★ — "RELOC-ONLY REMAP GATED DIFF" IS A CARVE VERDICT, NOT A CODEGEN VERDICT, UNTIL YOU DIFF THE `.text` (P31 S75; 4 siblings + 1 latent, ~1,300 ins) L34817 ### integration / TU plumbing (89) - **§8c** — Splitting a TU means rebuilding its DECLARATION ENVIRONMENT, not moving text (Phase 26 session 6) L456 - **§8d** — Templating a body INTO a TU must not CHANGE its declaration environment — demote the carried data externs (Phase 26 session 8, byte-proven on `func_8015AE2C` ×133) L502 - **§9.3** — Make it the build: resegment + swap + resolve (`tools/psyq_integrate.py`, libcd DONE) L704 - **§9.4** — Integrating a SECOND library (libgs block 6 after libcd) — multi-library gotchas L731 - **§9.5** — Integrating a WHOLE multi-block library in one call (full libgs — Phase 7 session G) L759 - **§17a** — The TOOLKIT at WAVE scale (Phase-18 Step-3b/Step-1 — measured) + the pipeline-integration gotchas L1468 - **§3-THE** — CAP: at this tail the match_one→gate gap is the LOOSE-TYPING CALL-GRAPH wall (not cheap plumbing) L1660 - **Diagnostic** — lesson: a failed in-TU build leaves a STALE `.o` L1677 - **§28d** — The "macro-extern-injection" lever: freeing reach-134 inline matches dedup_propagate skips as "not self-contained" (Phase 23, `tools/inject_capped_externs.py`) L2389 - **§30a** — §30 generalizes to the FULL near-miss backlog (via STANDARD Opus agents, not just Fable5) + 2 more steer levers + the mechanical-integration throughput unlock (Phase 23 (a)) L2417 - **§40a** — The DECL-RECONCILE pass (type-lift so remapped drafts compile in the sibling TU) (Phase 25 T7.2, 2026-07-08) L2624 - **§42a** — addendum — wave 2 (residuals + 29-100 band): iso-MATCH ≠ real-TU bank, the memcpy→struct-assign fix, +5 levers (2026-07-10b) L2991 - **§42c** — addendum — wave 3 (Max orchestrator + CORRECTED Ultracode fan-out, 2026-07-10c): the real-TU-faithful parallel harness (rtu_match) + 7/9 crack, ZERO iso-drift L3071 - **§56** — Banking a hand-drafted GIANT into its exemplar TU: self-contained draft vs live-TU decls, 4 reconciliation tactics (`func_8013FAF8`, 312 ins ×138, Phase 29 T4, 2026-07-17) L4225 - **§61** — Task 14: the gate ladder's missing stage is the ARITY pre-pass, and it is TU-side not draft-side (Phase 29, 2026-07-21) L4583 - **§61a** — The Task-5 wave: 11/12 MATCH, 0 banked — three DISTINCT integration walls, each now named (Phase 29, 2026-07-21) L4644 - **§63** — The fresh-138 DEF-SIDE blocker: fix the HEADER decl, not the draft (`fix_header_decl.py`, Phase 29 SESSION-13, 2026-07-23) L4915 - **§63** — UPDATE (Phase 29 SESSION-14) — `fix_header_decl`'s "SAFE" verdict is FLEET-BLIND; it MUST be R22-validated L5038 - **§65** — The stranded-draft recovery: BLAST-RADIUS TIERS, and the per-overlay de-macroize that refutes §20's DEF-conflict wall (`tools/demacroize.py` + `tools/blocker_probe.py`, Phase 29 SESSION-16, 2026-07-24) L5053 - **§65b** — The escape: de-macroize the instantiation, don't touch the shared header L5083 - **§65f** — The de-macroize lever's BOUNDARY, measured: byte-neutral 14 times of 15, and the 15th is why the gate exists L5156 - **§65g** — Where the cheap levers STOP: the local-type and in-TU-self-decl classes did not yield L5177 - **§85** — The RETURN-axis fleet widen is ALL-OR-NOTHING: widening the shared header alone guarantees a conflict in the source overlay (Phase 29 SESSION-20, `func_8012CC88` / `func_8014D12C`) L6577 - **§95** — `reconcile_tu` dropped the SIBLING declarators of a multi-symbol `extern` line (Phase 29 SESSION-21, `func_80176218`) L7021 - **§96** — The same rewrite, one shape down: `reconcile_tu` matched statements to lines by TEXT, so every COMMENTED declaration was silently skipped (Phase 29 SESSION-22, `func_80176218` banked) L7056 - **§3-Two** — `reconcile_tu` bugs found underneath, one introduced while fixing the other L7228 - **§100** — Prefer the DRAFT-LOCAL fix: a type only one function uses belongs in its BODY, not in a shared header (Phase 29 SESSION-22, `func_80175DA8` 0/137 → 137/137) L7254 - **§102** — A PLUMBING verdict can MASK a DIFF; and K&R is not always a codegen change (Phase 29 SESSION-22, `func_8016EC0C`) L7310 - **§103** — A FILE-scope `extern` in a shared overlay TU is a GLOBAL constraint on every LATER function; move the DECL, not the draft (Phase 29 T48/T51, `func_80135260` — the fleet-wide half) L7345 - **§109** — Conforming a definition to a shared header: fix the NAMES, then check the RETURN precondition (Phase 29 T60) L7638 - **§112** — A macro-scoped declaration only collides where the macro is INSTANTIATED (Phase 29 T67/T69, `audit_header_sigs.py`) L7768 - **§114** — The THIRD decl axis: a CALLEE the draft declares differently from the target TU (Phase 29 T76/T77) L7846 - **§3-The** — integration idioms (these decide whether a byte-correct draft BANKS) L8842 - **§3-The** — declaration surface (integration, not codegen) L8979 - **Reconciling** — a gate-refused draft: which way you edit depends on WHERE the TU's decl is L9603 - **§3-The** — same swallow, twice more, in the integration spine L9747 - **§3-A.** — Payload word0 is a global MODULE ID; code starts after the header L10565 - **§162** — THE LICM PAIR: what makes an address a movable AT ALL, and why the preheader order is the body order (P30 S47, `ov_MAIN_012`) L11181 - **§166** — THE DESTINATION-TU ORACLE (P30 S48): the seven-attempt bug that was never codegen L14960 - **§173** — THE STORED-PLUMBING RECOVERY RECIPE (P31 T6): symfix-first, per-group isolation, and where the verdicts have no drafts L16674 - **§176d** — THE CONFLICT TABLE MUST BE SEEDED FROM THE TU, AND KEYED PER FILE (P31 S52, 2026-08-15) L16871 - **§180b** — WHAT THE PRE-GATE LADDER ACTUALLY FINDS IN A COLD PILE (the shape of integration debt) L17887 - **§185** — EDIT THE SIDE THAT IS CHEAP TO VERIFY, AND CHECK A TU RETYPE AT ITS USE SITES L18105 - **§194-E** — The wave card's `exemplar` is the TARGET ITSELF on 42/73 wave-U and 36/71 wave-T cards — a construction consequence of `atlas.py:657` (exemplar = max-nins OPEN member) meeting `build_wave_atlas.py:166` (one card per gid); and the shipped §193-A `seed_ref` is same-binary on 0/51, so no card field can ever name a destination-TU sibling L19203 - **§195-J** — GTE / PsyQ op CALL-vs-INLINE is a PER-SITE SOURCE FACT, not a TU style — both forms coexist in one TU (measured in 2 TUs), and the tell is the target's own opcodes (`lwc2`/`sqr`/`swc2` vs `jal`), not the sibling. Costs -8 ins on func_8018505C. Corollary: the game's inline `sqr` macro emits TWO hazard nops, the SDK's `Square0` body emits ONE — so the inline form is provably not `Square0` inlined (a second instance of §187's SDK-vs-game GTE nop divergence). L20267 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§200** — THE ALIAS IS THE UNIVERSAL DECLARATION ESCAPE: stop negotiating with the TU's spelling (P31 S55) L21150 - **§201-A** — §150-B applies to `decl_prior`'s DEF row: for an overlay-window symbol the banked "definition" is usually another overlay's function, and the card ranks it ABOVE the destination TU L21209 - **§203** — A DEDUPED TYPEDEF MUST PRECEDE EVERY SPLICE POINT, NOT JUST ITS OWN (P31 S56) L21537 - **§210** — THE SINGLE-BIT MASK IN A BOOLEAN TAIL: `andi K ; sltu $zero,v` vs `srl n ; andi 1` is a STATEMENT-SHAPE dial, not an operator choice (P31 S58) L22379 - **ADD-2** — → §42a addendum — A SHARED CONSTANT *NAMED IN A LOCAL* ACROSS A `jal` IS AN ISO→TU DRIFT HAZARD; WRITE BARE LITERALS L25147 - **ADD-3** — → §236, item 10 — THE UN-DELETED `INCLUDE_ASM` STUB IS A DUPLICATE DEFINITION L25159 - **ADD-2** — → §236 addendum (item 1 corollary) — THE SILENT-SPLICE DECLARATION LADDER: WHERE THE TU HAS NO SPELLING, UNPROTOTYPED `()` BEATS THE FLEET VOTE L25377 - **§273** — A standalone compile is the WRONG oracle for a TU-destined draft (P31 S59) L25629 - **ADDENDUM** — to §134 — a typedef defined BELOW the splice point is stripped anyway L25697 - **ADDENDUM** — to §8 — WHEN `INCLUDE_RODATA` NEEDS A STANDALONE `.s` YOU CAN'T CREATE, CARRY THE FRAGMENT AS A FILE-SCOPE `__asm__` BLOB L26232 - **§280** — THE CURSOR'S DECLARED TYPE PICKS `sltu` vs `slt`: a `T*` bound test is UNSIGNED by C rule, an integer cursor is signed (P31 S60; wave cf, func_800CAE74, md_MAIN_031, byte-proven) L26966 - **§NNN** — A LOOP CURSOR'S C TYPE (POINTER vs PLAIN INTEGER) SELECTS `sltu` vs `slt` FOR ITS BOUND TEST, INDEPENDENT OF THE VALUES INVOLVED (P31 S60; `func_800CAE74`, md_MAIN_031, byte-proven; cross-confirmed same wave by `func_8017F2A4`, ov_SC03_096) L26967 - **ADDENDUM** — to §199-G — A `default:` LABEL GROUPED ONTO THE LAST CASE REMOVES THE `j default` TAIL, EVEN THOUGH THE 2-NODE HEADER STAYS ALL-POSITIVE L27421 - **§301** — AN INTERNAL `j` CARRIES `R_MIPS_26 .text`: rtu/match_one "MATCH" COULD NOT SEE WHICH LABEL A `j` TAKES — FIXED (`jrel`), AND THE TWO DRAFT SHAPES IT HID (P31 S62 T1; byte-proven 2/2, negative-controlled over 3,297 stubs) L29681 - **Addendum** — Reversed-operand sltu then xori 1 means unsigned less-or-equal (func_8017D948) L30515 - **§NNN** — REFUTED: A `sw %lo(SYM)($at)` IN A JUMP'S DELAY SLOT WITH ITS `lui $at` STRANDED ABOVE THE JUMP IS **ASSEMBLER** MACRO-EXPANSION, NOT A cse SPLIT — OUR PIPELINE CANNOT EMIT IT FROM ANY C, AND THE +1 IS A `maspsx` GAP (P31 S66 round2, `func_8005DBD8`, byte-probed) L30601 - **§320** — THE §43 "RETURN-TYPE FLIP PAIR" IS **NOT** TU-EDIT-REQUIRED: THREE DRAFT-ONLY ESCAPES (P31 S66; byte-proven func_800CCBC0, func_800D30D0, func_800D2A24) L30796 - **§323** — CARRYING FILE-SCOPE TYPES BETWEEN SPLIT TUs: DEDUPE BY NAME, SKIP HEADER-PROVIDED, AND THE GUARD TRAVELS WITH THE BLOCK (P31 S67; ov_SC02_000, two defects fixed, one open) L30906 - **§324** — THE ARITY-SELF WALL NEEDS *BOTH* HALVES: A K&R DEFINITION **AND** A NO-PROTO TU DECL; EITHER ALONE STILL FAILS (P31 S67; rtu-proven ov_SC01_005/func_8017FBCC, 14 instances measured) L30959 - **§337** — THE CC1-ONLY BLOCKER CLASS: `blocker_probe`'s STATIC ORACLE REPORTS "none" AND THE WHOLE-TU COMPILE STILL FAILS (P31 S67; ov_SC04_011/func_801827DC, md_MAIN_027/func_800CB4A4) L31167 - **§356** — MEASURE A DRAFT IN THE TU IT WILL LIVE IN, NOT IN THE STANDALONE PROBE (P31 S68; measured over 47 stored drafts) L31584 - **§362** — TWO TRAPS WHEN A CARVE MOVES A STUB INTO THE `-O0` TU (P31 S68; byte-proven, 6 fns / 2,547 ins across ov_MAIN_012 / ov_SC02_037 / ov_SC03_107) L31692 - **§367** — RECONCILING A DECL CONFLICT BETWEEN TWO DRAFTS FOR THE SAME TU (P31 S68; main) L31787 - **§376** — ★★★ — A STANDALONE `match_one` CLOSENESS OF 0 IS A CLAIM ABOUT THE **BODY**, NEVER ABOUT THE **TU** (P31 S69; measured 0/28) L32065 - **§378** — ★★★ — THE **SELF-CALLER CAST**: LET A TU KEEP CALLING THE FUNCTION IT IS ABOUT TO DEFINE (P31 S69; byte-proven ov_SC04_010/func_8017D6CC) L32124 - **§385** — ★★★ — THE **SCHED2 PRIORITY-DONOR ASM**: closing the "hoisted-invariant vs IV-init preheader swap" class (P31 S69; byte-proven main/func_80038A58, 347 ins, fable escalation 2 → 0) L32319 - **§378c** — ★★ — THE FIFTH DECL-BLOCKER VARIANT: THE DRAFT REDECLARES WHAT THE TU ALREADY OWNS (P31 S69) L32732 - **§398** — ★★★ — `family_remap` CARRIES THE **SOURCE** TU's DECL ENVIRONMENT INTO A DESTINATION THAT ALREADY OWNS THOSE NAMES (P31 S69; measured 3 banked of 22) L32845 - **§3-D.** — INTEGRATION IS STILL THE BOTTLENECK, AND THE TU IS THE AUTHORITY L33150 - **§415** — ★★ — A FILE-SCOPE DECL MAKES gcc-2.7.2 MERGE THE TU'S LATER *BLOCK-SCOPE* EXTERNS INTO IT (P31 S71; byte-proven `ov_SC04_011/func_80180B24`, 215 ins) L33557 - **§423** — ★★★ — "MATCH IN ISOLATION + GATE REJECTS + CAUSE NOT DETERMINED" ⇒ GREP THE TU FOR A FILE-SCOPE TYPEDEF THE DRAFT ALSO CARRIES (P31 S71; byte-proven `ov_SC03_092/func_8017FA74`) L33790 - **§431** — ★★★ — SPLITTING A 27,000-LINE TU AT ITS ORIGINAL BOUNDARIES: THE JTBL SPANS TELL YOU WHERE, AND THE COMPILER TELLS YOU WHAT CROSSES (P31 S72; `src/800.c` -> `800.c`/`800_b.c`/`800_c.c`, byte-identical with nothing banked) L34132 - **§435** — ★★★ — AN OVERLAY TU SPLIT IS NEAR-FREE, AND THE GAP TEST IS "IS THIS WORD A CODE ADDRESS", NOT "IS IT ZERO" (P31 S74; four overlays split, all byte-identical, CARVE-BLOCKED class emptied fleet-wide) L34317 - **§437** — ★★★ — A WRAPPED TRAILING COMMENT BROKE THE TU PARSER, AND THE §8b DECL LAYER WROTE `extern #define` (P31 S74; five independently-MATCHed jr bodies unblocked, `ov_SC06_029` byte-identical) L34431 - **§440** — ★★★ — A `.rodata` CARVE PIECE BINDS TO A SUBSEG, NOT TO A FUNCTION: EXTEND THE CARVE INSTEAD OF ISOLATING (P31 S74; 6 banked, resident included) L34586 - **§442** — ★★★ — A RECOVERY RUNG THAT REWRITES A LEGAL CONSTRUCT INTO AN ILLEGAL ONE READS EXACTLY LIKE A CODEGEN WALL (P31 S75; `reconcile_tu`, 344 ins unblocked) L34657 - **§477** — ★★★ — THE `self_decl_tu` CLASS IS A SOLVED, MECHANICAL LANE: 16 DRAFTS, 16 BANKS (P31 S77) L35723 - **§492** — ★★ — "C-PLUMBING" WAS THREE DIFFERENT THINGS (P31 S79 #7): A RAW SPLICE THE GATE'S LADDER BROKE, TWO -O0 BODIES THE CHECKER COMPILED AT -O2, AND TWO DRAFTS THAT BELONGED TO OTHER OVERLAYS L36402 - **§493** — ★★ — THE PERMUTER ROUTE END-TO-END, AND THE THREE PLUMBING STEPS BETWEEN A SCORE-0 WINNER AND THE MAIN GATE (P31 S79 #8) L36430 - **§494** — ★★★ — TEN BANKS FROM ONE-AGENT-PER-FUNCTION DRAFTING (P31 S79/S80 #9): THE IDIOMS, THE PLUMBING, AND THE THREE WAYS AN AGENT'S "MATCH" WAS NOT ONE L36490 - **§496** — ★★ — A CARRIED-TYPE TEST THAT ASSUMES THE OVERLAY INCLUDE SET SILENTLY DROPS A RESIDENT TYPEDEF (P32 T1a; byte-proven, resident `func_800D128C` isolation) L36671 - **§453** — ★★★ — ONE SOURCE PER UNIQUE FUNCTION: THE INCLUDE-AT-SITE SHARE, ITS GATE, AND THE FOUR DEFECTS THAT SHAPED IT (P35 S94–S96) L37561 - **§457** — The lever-removal move catalog, S105: eight TU batches, 42 of 42 at 0 (Phase 36 T7) L37814 ### build graph, splat & the harness (219) - **§4** — Flag/toolchain gotchas L190 - **Build** — mechanism — per-file opt override (splat resegmentation) L307 - **§8c** — Splitting a TU means rebuilding its DECLARATION ENVIRONMENT, not moving text (Phase 26 session 6) L456 - **§9** — Link real PsyQ library objects byte-exact (Phase 7 — GO proven) L632 - **§9.1** — Generalised per-object linker — `tools/psyq_link.py` (+ `psyq_link_lib.py`), 18/18 libcd byte-exact L661 - **§9.2** — Wire a library region into the build with NOLOAD — no data carving (`tools/psyq_link_region.py`) L685 - **§9.3** — Make it the build: resegment + swap + resolve (`tools/psyq_integrate.py`, libcd DONE) L704 - **§9.6** — Scaling library linking to the whole EXE (Phase 8 — 8 libs linked, 20%→50% byte-identical) L785 - **§9.7** — Binary-agnostic toolchain refactor (Phase 9) — the reusable pattern for Gen2 L821 - **§3-The** — clean object-level metric (use this, not the permuter score, for jtbl/rodata functions) L863 - **§3-The** — mechanism: game-code dedup is SOURCE-LEVEL, not an object swap (R-D1, the key lesson) L945 - **Per-binary** — toolchain provenance (R24) L978 - **§12** — Ultracode harvest — parallel-draft + byte-gate at scale (Phase 12, resident: 1.4%→71.7% in one session) L983 - **One-command** — onboarding — `tools/new_overlay.sh ` L1081 - **§3-THE** — NON-4-ALIGNED-OVERLAY GOTCHA (≈75% of the fleet; fixed in the template + Makefile, automatic) L1098 - **Fleet** — build — `make build-all` / `make check-all` L1139 - **§14** — Propagate a matched function across the fleet — `tools/dedup_propagate.py` (Phase 15) L1145 - **§14b** — Harvesting the UNMATCHED shared core — the match_one wall (Phase 15) L1215 - **§14c** — Callee-signature-aware harvest — breaking the extern-type-conflict wall (Phase 15, T6) L1236 - **§17** — The compiler-quirk wall — the matching TOOLKIT (Phase 18; gcc-2.7.2 source + byte-gated) L1399 - **§3-The** — pipeline gotcha — match_one ≠ the gate; `sig_unify` is MANDATORY L1444 - **§17a** — The TOOLKIT at WAVE scale (Phase-18 Step-3b/Step-1 — measured) + the pipeline-integration gotchas L1468 - **§19** — Scaling the toolkit waves — the recovery PIPELINE + the propagation CAP (Phase 19 T3) L1620 - **§20** — The wave-at-scale GATE CAP + residual-class verdicts (Phase 20) L1655 - **§3-THE** — CAP: at this tail the match_one→gate gap is the LOOSE-TYPING CALL-GRAPH wall (not cheap plumbing) L1660 - **Diagnostic** — lesson: a failed in-TU build leaves a STALE `.o` L1677 - **Phase-20** — RESOLUTION — `tools/cast_call_sites.py` BUILT + the cap re-diagnosed (R14, byte-proven) L1724 - **§24** — The `ov_SC01_077_a.c` split-file vein: split-aware propagation, but loose-typing-gated (Phase 21) L2108 - **§25** — The "schedule" class is mostly COALESCING (pin-crackable), not scheduling; + the gate two-stage + h_exact over-counts ×134 (Phase 21, cont.6) L2155 - **§3-The** — gate two-stage — sig_unify is a FALLBACK, not unconditional (`gate_stage.py`, §19 folded in) L2212 - **§28** — Banking a "close=0 gate-rejected" giant: the canonical-extern recovery (Phase 22 T1, byte-proven on func_8015126C ×134) L2348 - **§28c** — The close=0 recovery is NOT fully exhausted (§26 corrected, R14); + the dedup_propagate registry-skip recovery (Phase 22 T2) L2375 - **§28d** — The "macro-extern-injection" lever: freeing reach-134 inline matches dedup_propagate skips as "not self-contained" (Phase 23, `tools/inject_capped_externs.py`) L2389 - **§29** — Reasoning-model (GLM5.2) DEF-side reconciliation idioms + the wall's hard limit (Phase 23 T10.7, `tools/glm_reconcile.py`) L2397 - **§33** — Automating the giant decl-reconcile: `tools/reconcile_decls.py` (the DATA analog of §20's `cast_call_sites`) + a fleet-majority type oracle (Phase 24 T7b, byte-proven on `func_80129CF8`) L2457 - **§34** — The `func_80138ED0` giant crack: gcc-2.7.2's **3-qty sort bug** + the **zero-byte asm allocation toolkit** + the **giv-init fence** (Phase 24 T5; Opus→close=21, Fable5→MATCH ×134) L2477 - **§35** — The region-a sibling-giant harvest: difficulty ≠ $s-reg count (it's global-array hoisting) + the banking recipe + new loop idioms (Phase 24 T7, 5 parallel Opus-Max agents) L2494 - **§37** — The T7 §G giant endgame: 6/8 cracked, meta-laws + transferable levers (Phase 24, 2026-07-07; FULL byte-verified detail + gcc-2.7.2 line cites in `docs/gcc-2.7.2-map/t7g-giant-harvest.md`) L2529 - **§39** — The ×1→×134 giant-endgame: propagate a matched **-O2** giant via the NATIVE DEFINE-macro path (Phase 24 T7 §G close, 2026-07-08) L2560 - **§41** — The DEF-SIDE canonical-sig wall: mechanically banking a drafted giant past `conflicting types` (Phase 25 T5b batch-2, 2026-07-09; `tools/canon_sig_reconcile.py`, byte-proven on `func_8013B274`) L2733 - **§41a** — v3.1: the def-side wall was ~71% TOOL-shaped — the five measured defects + the laws that dissolve them (Phase 25 T6, Fable5, 2026-07-09) L2792 - **§41b** — T7 execution: the object-only probe OVER-counts BANKABLE by two link/rodata classes (Phase 25 T7-M1, 2026-07-10) L2847 - **§41d** — `void`→`s32` is NOT always byte-neutral: gate the RAW draft FIRST (Phase 26 session 6, byte-proven) L2916 - **§42b** — addendum — wave 3 (Max, 2026-07-10c): THE STALE-OBJECT GATE TRAP + the read-global `&`-cast drift + fix L3032 - **§45** — The flagship `func_80133CD4` crack (399 ins ×134): the merged-variable permutation-breaker + the 1-death local-alloc gate (Phase 25 task A giant escalation, Fable5 gdb-on-cc1, 2026-07-11) L3336 - **§51** — TOOLING INTEGRITY: the silent skip, and how to hunt it L3668 - **§51b** — Why the byte-gate cannot save you L3690 - **§51f** — Checklist for any new corpus-scanning tool L3793 - **§51g** — When the thing you are scanning has a GRAMMAR, parse the grammar (`tools/cdecl.py`) L3807 - **§53** — SWEEP A FAMILY WITH THE TOOL ITS EXEMPLAR NEEDED: the jr/switch carve, and how omitting it manufactured the "families don't template" doctrine (Phase 28 T1, 2026-07-15) L4066 - **§55** — Core-crack wave levers + the GATE-ORCHESTRATION law (Phase 29 T3, 13-agent ultracode wave, 2026-07-17) L4131 - **§55b** — THE GATE-ORCHESTRATION LAW (3 traps, ~3.5h lost; all recovered, 0 data lost) L4160 - **§55c** — Sizing the propagate: a TARGETED propagate is ~4 min/core, and "it's slow" was a BROKEN-TREE ARTIFACT L4179 - **§54** — `--fix-def-sig`: the member's CANONICAL DECLARATION is a build step too (tiny-IMM mega-pools, +4,801, Phase 29 T6, 2026-07-16) L4201 - **§57** — The SELF-decl normalize: the sibling's OWN caller declares the templated fn divergently (the third §17a-1 direction; `tools/normalize_self_decls.py`, func_801670E4 ×137, Phase 29, 2026-07-18) L4294 - **§61** — Task 14: the gate ladder's missing stage is the ARITY pre-pass, and it is TU-side not draft-side (Phase 29, 2026-07-21) L4583 - **§61b** — The jtbl gate stage: built, and the ORDERING law it exposed (Phase 29 Task-14 stage 4, 2026-07-21) L4696 - **§61d** — The undo was eating the tree: two tools, one defect, invisible to the byte-gate (Phase 29, 2026-07-22) L4817 - **§62** — The jtbl RECONCILE must also follow the carve: the post-carve draft reconcile (`harvest_verify._jtbl_reconcile`, Phase 29 SESSION-11, 2026-07-22) L4871 - **§65** — The stranded-draft recovery: BLAST-RADIUS TIERS, and the per-overlay de-macroize that refutes §20's DEF-conflict wall (`tools/demacroize.py` + `tools/blocker_probe.py`, Phase 29 SESSION-16, 2026-07-24) L5053 - **§65d** — Existing-ladder baseline, measured (do this before building a recovery stage) L5129 - **§65e** — Two oracles, and the disagreement is the finding (R34 in practice) L5139 - **§65f** — The de-macroize lever's BOUNDARY, measured: byte-neutral 14 times of 15, and the 15th is why the gate exists L5156 - **§66b** — A metric parsed out of another tool's prose goes NULL silently when the label changes L5238 - **§67a** — Run the symbol-set guard BEFORE you pay for a gate (`tools/symcheck.py`, Phase 29 SESSION-18) L5443 - **§77** — Every extraction tool carries a NARROW hard-coded preamble set; anything outside it silently caps the body's reach. Diff the preamble before you gate. (Phase 29 SESSION-19 — three variants in one session, two different tools) L6122 - **§3-The** — CANDIDATE gate and the REAL gate need DIFFERENT preambles — keep the difference out of the bank L6162 - **§79** — For a 0-callee giant, fingerprint by DATA symbols (§71 cannot fire); and the STACK-SLOT ORDER is a declaration-order oracle (Phase 29 SESSION-19, `func_8017BF14` 4,763 ins, cold start → 45/4763) L6258 - **§82** — Two source-shape oracles from behemoth #6: a duplicated `addiu $aN,$sp,K` across a `jal` means the block was INLINED, and scalar-vs-aggregate decides WHEN a stack slot is allocated (Phase 29 SESSION-19, `func_8017C730` 1,061 ins) L6408 - **§3-2.** — Scalar vs aggregate decides *when* the slot is allocated L6427 - **§3-Why** — it survived every candidate gate L6533 - **§3-THE** — FIX IS MECHANICAL — the tool already holds the answer L6548 - **§87** — `match_one` COMPILES but never LINKS, so an unresolvable data symbol reads as MATCH; and stored drafts go STALE against the tree (Phase 29 SESSION-20) L6671 - **§88f** — the missing rung: a RELOCATION gate between `match_one` and the binary L6756 - **§89a** — MEASURE the write set; do not assert its tier (`tools/blast_radius.py`) L6770 - **§89b** — the parallel gate farm existed; the family path could not reach it (`tools/sweep_parallel.py`) L6783 - **§90** — Five tool-integrity laws from one session, each of which changed an answer (Phase 29 SESSION-21, 2026-07-27) L6804 - **§90a** — A comparison tool MUST share its reference oracle's index space, exactly L6810 - **§92** — Conforming a DECLARATION: pointer changes are caller-neutral, scalar-WIDTH changes are not (Phase 29 SESSION-21, `tools/conform_decls.py`) L6929 - **§97** — The gate's own tree hygiene: a refused carve, an unchecked recovery, and a snapshot that captured a dirty tree (Phase 29 SESSION-22) L7104 - **§105** — A gate's revert must survive an EXCEPTION, not just a failure (Phase 29 T53, `jtbl_family_bank`) L7469 - **§107** — A lever wired into ONE gate path is a lever most families cannot reach (Phase 29 T56, `func_80144090` 0/136 → 136/136) L7551 - **§3-Why** — it survived so long: a MASKED oracle will MATCH a wrong symbol L7991 - **§122** — GATE RAW BEFORE TRANSFORMING; the undo belongs to the WRITER, as a per-edit journal (P30 T0a, 2026-07-30) L8122 - **§123** — PROPAGATE A FAMILY WITH THE TOOL ITS TIER NEEDS: `dedup_propagate` is h_exact-only; its refusals are statements about the TOOL (P30 wave 1, 2026-07-30) L8154 - **§126a** — a bare `except: continue` around a coverage-asserting oracle re-creates the silent skip (P30 S28) L8355 - **§130** — An INCREMENTAL build can report BYTE-IDENTICAL for a change the CLEAN build cannot even LINK (P30 S28, the jr pair) L8524 - **§132** — The `JR-PAIR-IN-ONE-O0-OBJECT` "wall" was TWO instrument defects: a merged-double span the carve could not see, and a truncated object no rule deleted (P30 S29, `func_8013B83C` + `func_8013BD74`) L8608 - **§133** — The DEFAULT-FILTER class: three times in one session, a tool silently answered a narrower question than the one asked (P30 S1–S3) L8723 - **§134** — MULTI-LINE BLINDNESS: one root cause, four faces, in `family_remap`'s preamble scanner (P30 S6, 190 zero-crack families) L8749 - **§3-The** — splitting/merging rules (each closed a residual, byte-gated) L8891 - **§137a** — A gate verdict has a TIMESTAMP; re-check it against the draft's mtime L9424 - **§138** — The propagation lanes: a gate refusal is a DECLARATION, and which lever you owe depends on blast radius L9453 - **§134** — again, in a second tool — and the waiter rule corrected L9564 - **Reconciling** — a gate-refused draft: which way you edit depends on WHERE the TU's decl is L9603 - **§139** — A GATE THAT GREPS FOR VERDICTS MUST ASSERT 1:1 ACCOUNTING; and a `--src` filter must not survive a carve (P30 S38, wave 6: 10 of 16 drafts vanished) L9632 - **§140** — A METRIC IS NOT A MEASUREMENT UNTIL IT IS REPRODUCIBLE FROM THE COMMITTED TREE (P30 S1e: a phantom regression that gated the session's best lever) L9698 - **§142** — An open stub whose `h_exact` class is MATCHED elsewhere is FREE. Propagate the body; do not gate a draft. (P30 S39, +7,710 ins in two commands) L9816 - **§3-The** — measurement (do this before any wave; it is ~20 lines and needs no builds) L9828 - **§155b** — check the TYPE your oracle returns before comparing against it (S45 p5) L10625 - **§156** — an ORPHANED reconcile poisons the fleet: `dedup_propagate`'s kept edit (S45 p6/p7) L10678 - **§158** — The RANGE-EXTENDER: a fourth zero-emission asm lever completes the allocno toolkit (P30 S46 tier-3, `func_8017CE58`, 733 ins) L10750 - **§160** — THE REACH-15 WAVE HARVEST: an align-1 block move, and the instrument that called it a wall (P30 S47) L10931 - **§161** — THE RETRY-WAVE HARVEST: a jump table indexed from zero, and two allocator traps (P30 S47) L11007 - **§162** — S48 WAVE-1 HARVEST (P30, 2026-08-11): the reach-ordered sibling campaign's first 12 targets L11052 - **§163** — S48 WAVES 2-3 HARVEST (P30, 2026-08-11/12): the five that were byte-probed and are actionable L11798 - **§165** — S48 WAVE-4 HARVEST (P30, 2026-08-12): banked the same day the wave landed L13740 - **§166** — THE DESTINATION-TU ORACLE (P30 S48): the seven-attempt bug that was never codegen L14960 - **§167** — S48 WAVE-5/6 HARVEST (P30, 2026-08-12): the saturation point L15016 - **§176e** — SYMBOL IDENTITY IS COMPUTABLE OFFLINE (P31 S52): `tools/reloc_identity.py` L16910 - **§3-A.** — The seven under-reporting holes (all in `gate_main`, all the same shape) L17044 - **§3-C2.** — RECONCILE BEFORE THE FIRST GATE — a parked draft gets HARDER to bank, not easier L17091 - **§3-D.** — The measured cost shape, and what to build next L17114 - **§176i** — WHAT A STATIC PRE-GATE CHECK CAN AND CANNOT PROVE (P31 S52, wave Q) L17132 - **§179** — IDIOMS MINED FROM THE WAVE P/Q JOURNALS (P31 S52, harvest pass) L17371 - **§179-G** — 🟡 A PIN CAN **CREATE** A COMBINE `LOG_LINK` AND DELETE AN `andi` (sixth RC-5 channel, n=1) L17600 - **§176c** — MAIN (SLUS_007.26) CANNOT BE GATED INCREMENTALLY L17667 - **§180** — THE LEFTOVER-DRAFT HARVEST: RE-VERIFY WHAT YOU ALREADY HAVE BEFORE DRAFTING ANYTHING NEW L17858 - **§180b** — WHAT THE PRE-GATE LADDER ACTUALLY FINDS IN A COLD PILE (the shape of integration debt) L17887 - **§181** — WHAT A WAVE'S GATE ACTUALLY REJECTS (P31 S53, measured on wave R's 45-draft main pile) L17924 - **§184** — COMMENT-BLINDNESS IS A DEFECT CLASS, NOT A BUG (P31 S53: three tools, one root cause, one session) L18077 - **§187** — 🔴 "SAME SOURCE" IS NOT "SAME OBJECT": THE SDK BUILD AND THE GAME BUILD DISAGREE ON GTE NOPS L18183 - **§190** — THREE PRESCRIPTIONS FROM THE SAME HARVEST (weaker evidence than §189, honestly labelled) L18346 - **§191** — WHAT THIS HARVEST DID **NOT** BANK (4 rejected, 4 narrowed) — recorded so it is not re-derived L18383 - **§192** — THE PRE-GATE LADDER WAS MAIN-ONLY, AND NOBODY COULD SEE IT (P31 S54) L18405 - **§193** — THE WAVE-T HARVEST (P31 S54): 71 index_gap reports -> 9 laws, 5 rejected, 61 already-covered L18470 - **§193-A** — The wave card ships only OPEN-set pointers and discards the atlas's BANKED one — `exemplar`/`sibs` are stubs 100% by construction (tools/atlas.py:96/657), while `seed.ref` (matched pool, atlas.py:505-536) is dropped at build_wave_atlas.py:143 L18487 - **§193-G** — §164-54's "scope to ≥4 arms" bound is byte-wrong — the dispatch-topology oracle goes live at THREE case nodes (`balance_case_nodes` splits at `i > 2`), but only for a signed-after-promotion index L18791 - **§193-I** — A DECLARED AGGREGATE LOCAL HAS AN 8-BYTE FRAME STRIDE — CEIL(size,8), NOT size. N ARRAYS THEREFORE COST 8·k MORE THAN ONE HAND-LAID STRUCT (k = how many of them have size mod 8 ≠ 0), AND THE TELL IS IDENTICAL INSTRUCTION COUNT WITH EVERY $sp DISPLACEMENT SHIFTED BY THE SAME CONSTANT. L18892 - **§193-REJECTED** — what this harvest did NOT bank (recorded so it is not re-derived) L18951 - **§194** — THE WAVE-U HARVEST (P31 S54): 64 index_gap reports -> 14 laws, 5 rejected, 44 already-covered L18978 - **§194-E** — The wave card's `exemplar` is the TARGET ITSELF on 42/73 wave-U and 36/71 wave-T cards — a construction consequence of `atlas.py:657` (exemplar = max-nins OPEN member) meeting `build_wave_atlas.py:166` (one card per gid); and the shipped §193-A `seed_ref` is same-binary on 0/51, so no card field can ever name a destination-TU sibling L19203 - **§194-G** — Reading the integer half of a 16.16 stack aggregate: REGISTER-LIVENESS, not the C spelling, decides `sra` vs `lhu` — and the break is count-neutral L19313 - **§194-K** — Blind sched1's alias oracle with a second SET of a pointer pseudo — the first zero-byte, non-volatile, dependence-CREATING lever (corrects §167-05's "volatile is the only door"; fourth consumer of reg_n_sets) L19489 - **§194-REJECTED** — what this harvest did NOT bank (recorded so it is not re-derived) L19733 - **§195** — THE WAVE-V HARVEST (P31 S54): 67 index_gap reports -> 14 laws, 9 rejected, 76 already-covered L19754 - **§195-I** — §145(b) AMENDED — the pointer-bump `addiu` is saved by cse, not combine: ANY second SET of a pseudo in the address's equivalence chain (in-place `p += K`, §145(b)'s `p = r;` copy, or an asm re-tie) kills the fold; the pass is cse and the gate is `invalidate`'s `reg_tick++` L20202 - **§195-M** — Frame `vars` is a SEQUENTIAL bump-allocation, not a flat sum: §193-I's CEIL(aggregate,8) term and §165-03/§167-06's 8×orphan term are the SAME frame_offset walk at two different compiler stages, and each stage re-CEILs frame_offset to 8 before it allocates L20415 - **§195-REJECTED** — what this harvest did NOT bank (recorded so it is not re-derived) L20511 - **§197** — THE WAVE-W HARVEST (P31 S54): 68 index_gap reports -> 4 laws, 3 rejected, 41 already-covered L20612 - **§199** — THE WAVE-X HARVEST (P31 S54/S55): 63 index_gap reports -> 7 laws, 2 rejected, 56 already-covered L20758 - **§199-C** — A NEGATIVE CONSTANT MULTIPLY ALWAYS TAKES expmed's negate_variant — but whether you ever SEE the neg is decided by COMBINE, and for an EVEN |K| it never disappears L20863 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§199-G** — At TWO case nodes the switch-vs-if oracle is not blind — but the tell is ALL tests positive + a trailing `j default`, NOT the first test's polarity L21084 - **§201** — THE WAVE-Y HARVEST (P31 S55): 67 gap reports -> 5 laws, 8 rejected, 53 already-covered L21199 - **§204** — THE WAVE-Z HARVEST (P31 S56): 82 gap reports -> 5 laws, 16 rejected, 30 already-covered L21598 - **§207** — THE WAVE ab–ag HARVEST (P31 S58): 278 byte-banked notes → 25 laws, 103 self-reported no-gap L22232 - **§233** — THE WAVE aa–bg HARVEST (P31 S58b): 1,101 byte-banked notes → 24 new laws, 21 addenda, ~700 already-covered L23323 - **§236** — THE DECLARATION LAYER IS THE DOMINANT BANK-BLOCKER: NINE WAYS A BYTE-PERFECT BODY FAILS THE GATE (P31 S58b) L23437 - **§247** — TWO BRANCHES TO **ONE** LABEL MEANS THE SOURCE CONDITION IS NEGATED (P31 S58b) L23883 - **§254** — THE DEAD PARAMETER IS A REGISTER-PLACEMENT TOOL **(single observation — not yet cross-confirmed)** (P31 S58b) L24077 - **§259** — THE DISCARD LEDGER FOR THE aa–bg HARVEST (P31 S58b): WHAT WAS MINED AND REJECTED, AND WHY L24734 - **§260-A** — STAGE 2 IS PROVEN, AND THE WHOLE jtbl PIPELINE IS AUTOMATED AT THE GATE (P31 S59, same day) L24823 - **§261** — THE -O0 ORACLE: DERIVE THE OPT LEVEL FROM THE TARGET, AND `$fp` IS NOT THE TELL (P31 S59) L24859 - **§267** — ADDENDA HARVESTED FROM WAVES at/bh/bk/bl (P31 S59b) L25130 - **ADD-9** — → §213-3 / §217 addendum — ADJACENT PRE-CALL FRAME STORES ARE ONE AGGREGATE; THE UNESCAPED SCALAR NEIGHBOR IS DEAD-STORED L25225 - **§269** — ADDENDA HARVESTED FROM WAVES ax/bm (P31 S59c) L25349 - **§271** — Ordinal IMM pairing: text order is NOT emission order; emit CANDIDATES, let the oracle pick (P31 S59) L25587 - **§273** — A standalone compile is the WRONG oracle for a TU-destined draft (P31 S59) L25629 - **§274** — ADDENDA HARVESTED FROM 18 WAVES (P31 S60): 315 candidates, 255 already covered, 21 sharpenings, 3 new laws L25644 - **§278** — ADDENDA HARVESTED FROM WAVE cf (P31 S60): 34 candidates, 13 already covered, 8 sharpenings, 4 new laws L26635 - **§283** — ADDENDA HARVESTED FROM THE 36-WAVE BATCH (P31 S60): 1,216 candidates, 859 already covered, 46 sharpenings, 9 new laws L27131 - **§287** — A STACK-FRAME HOLE BELOW TWO ADDRESS-TAKEN AGGREGATE LOCALS IS A LEADING PAD MEMBER OF ONE COMBINED STRUCT, NOT SEPARATE LOCALS OR A REGISTER PIN (P31 S60/dj; `func_8017D104`, ov_SC06_027, byte-proven 47/47) (P31 S60; waves #, byte-proven) L29096 - **§293** — THE "WHOLE-OBJECT GATE NEEDS EVERY SIBLING" LAW, WRITTEN PROPERLY: IT IS THREE LAWS, AND THE LOAD-BEARING ONE IS "THE BASELINE, NOT THE SIBLINGS" (P31 S61; byte-proven on 15 binaries in one night) L29119 - **§294** — ADDENDA HARVESTED FROM THE S61 GEN0/MAXTOK/MAIN/DEEPSEEK BATCH (waves ab8/ab16/ab24/ab32 · g0a–g0f · m0a · ds1/ds2): 99 candidates, 52 covered, 15 notes → 10 addenda, 29 notes → 5 new laws (§295–§299), 3 refuted L29169 - **§303** — MODULE ISLAND TABLES: DERIVE THE PADS AT BUILD TIME, PEEL NOTHING — THE §154-A/§260 "island-pads"/"island-blocked" WALLS DISSOLVE (P31 S62 T3a; byte-proven md_SC03_076 func_801F0A9C + func_801F0F28, sha 9a165e36…) L29773 - **§305** — "CARVE-REFUSED" AT GATE TIME IS THREE NAMED, DETERMINISTIC CLASSES — NONE OF THEM A CARVE (P31 S62 T3; 28 resolver drafts autopsied 28/28) L29832 - **§313** — A RECOVERY STAGE THAT REWRITES THE DRAFT IS PART OF THE MEASUREMENT: GATE THE RAW DRAFT FIRST, THE REWRITE ONLY AS A FALLBACK (P31 S65 t5r-recovery; byte-proven func_8017BEBC, 246 ins) L30195 - **§314** — AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES ALL COMPILE BYTE-IDENTICAL, AND NONE REACHES A TARGET THAT SINKS THE NEGATIVE ARM INTO AN OUT-OF-LINE BLOCK (P31 S65; UNPROVEN, gate-refused draft func_80181720) L30251 - **§315** — ALL-CONSTANT AGGREGATE FILL: THE EMISSION ORDER IS SHARED-LITERAL GROUPS × DESCENDING INDEX, WITH PAIRED SUB-FIELDS INTERLEAVED (P31 S66; ⚠ **UNPROVEN** — `func_80185214`, gate-REFUSED draft, closeness 32→5, never MATCHed) L30283 - **§316** — A GUARD CHAIN THAT GATES A SUCCESS BLOCK ROUTES ITS FAILING GUARDS TO A LABEL **BEFORE** THE FLAG TEST, NOT TO THE EPILOGUE: THE RESIDUAL IS IMM-OFFSET-ONLY AT AN EXACTLY EQUAL INSTRUCTION COUNT (P31 S66; byte-proven func_80180284) L30320 - **§318** — A UNARY MINUS STORED BACK INTO THE SAME HALFWORD IS COMPUTED IN HImode, SO ITS LOAD IS `lhu`; AN INLINE `(s32)` CAST IS PROVABLY INERT AND ONLY A NAMED SImode LOCAL BUYS `lh` (P31 S66; `func_8018568C`, ov_SC03_028 — **`match_one`-only, NOT byte-gated**) L30388 - **Addendum** — Addendum to §167-12 — the single-operand volatile keepalive, re-confirmed at a gated MATCH (func_8017F498) L30495 - **§NNN** — REFUTED: A CODE BLOCK SHARED BY TWO *DIFFERENT* SWITCH STATEMENTS IS ORDINARY FORWARD `cross_jump`, NOT A SHAPE C CANNOT REACH — §164-11 ALREADY BANKS IT FROM PLAIN NESTED SWITCHES (P31 S66 round3, `func_8001B0D4`; ⚠ **UNPROVEN** — gate-REFUSED draft, closeness 68→45, never MATCHed) L30650 - **§323a** — R53, TWICE IN ONE HOUR: A FAILED BUILD LEAVES THE PREVIOUS BINARY, AND `sha1sum` READS GREEN (P31 S67) L30936 - **§328** — THE VOLATILE ALIAS MUST BE AN *OBJECT*, NOT A CAST: `*(volatile s32*)&sym` UNFOLDS %lo INTO A SEPARATE `addiu` (P31 S67; byte-proven ov_SC07_007/func_80181B8C, 156 ins, NEW LAW) L31034 - **§332a** — MAIN'S RESIDUAL FRONTIER IS CONTAMINATED WITH TOOLCHAIN WALLS: 4 OF 7 IN ONE WAVE (P31 S67, measured) L31106 - **§333** — FRAME SIZE IS SET BY *DECLARED* AGGREGATES, NOT USED ONES: AN UNREFERENCED TRAILING LOCAL IS A REAL DIAL (P31 S67; three independent byte-proven instances in one wave) L31125 - **§337** — THE CC1-ONLY BLOCKER CLASS: `blocker_probe`'s STATIC ORACLE REPORTS "none" AND THE WHOLE-TU COMPILE STILL FAILS (P31 S67; ov_SC04_011/func_801827DC, md_MAIN_027/func_800CB4A4) L31167 - **§340** — §194-K COROLLARY: **FLIP THE FALSE EDGE YOU CANNOT DELETE.** A "scheduler" residual can be sched.c's ALIAS ORACLE emitting a FALSE true-dependence; source order chooses its DIRECTION (P31 S67; byte-proven ov_SC03_107/func_8017CF48, 10 -> 0 in one compile, zero bytes) L31219 - **§346** — `c ? X : -X` TAKES expand_expr's COND_EXPR **SINGLETON** PATH (copy, then negate IN PLACE) — AN if/else STATEMENT GIVES THE TWO-ARM FORM (P31 S67; byte-proven ov_SC03_102/func_80180C38, closed the last instruction) L31332 - **§353** — USE `-fno-thread-jumps` AS AN **ORACLE** TO PROVE A RESIDUAL IS thread_jumps, THEN LAUNDER THE *VALUE* TO KEEP A DEAD RE-TEST (P31 S67; byte-proven ov_SC01_008/func_8017EC68, 279 ins) L31529 - **§357** — ONE STRUCT POINTER, NOT TWO: A SECOND SOURCE VARIABLE BUILDS A THIRD IV (P31 S68; byte-proven ov_SC06_029/func_80181DF8, 335 ins, 330 → 13) L31599 - **§358** — (sharpens §333) — AN **UNREFERENCED** FIXED-SIZE AGGREGATE LOCAL IS LOAD-BEARING (P31 S68; same function) L31608 - **§371** — ★★ — CARVING A **SINGLE-OBJECT MODULE BINARY**, AND THE spimdisasm RODATA-MIGRATION TRAP THAT FOLLOWS (P31 S68; byte-proven md_MAIN_003, func_800D0D6C 345 ins) L31900 - **§383** — TWO TOOLCHAIN FACTS THE PACKS DID NOT CARRY (P31 S69) L32254 - **§388** — ★★★ — THE **-O0 COLOURING ORACLE**: simulate `stupid.c` instead of grinding spellings (P31 S69; main/func_80011380 proved a C-level WALL at 6) L32394 - **§392** — ★★★ — THE SONNET-WAVE HARVEST (P31 S69): SEVEN TYPE/ORDER LEVERS THAT EACH CLOSED A MATCH L32512 - **§332b** — ★★★ — THE §332 "WALLS" ARE A PER-OBJECT ASSEMBLER MODE, NOT A C LIMIT — 6 CLOSE AS REAL C (P31 S69, Fable-3) L32712 - **§405** — ★★★ — THE S70 WAVE HARVEST: 130 agents, 113 MATCH, and the laws they brought back L33102 - **§3-A.** — THE ORACLE HAS A HOLE: match_one cannot see a jump table (§405-A) L33108 - **§414** — ★★★ — `parallel_gate` ON `main` IS A FALSE PASS, AND THE RULE WAS ALREADY WRITTEN DOWN (P31 S71) L33520 - **§423** — ★★★ — "MATCH IN ISOLATION + GATE REJECTS + CAUSE NOT DETERMINED" ⇒ GREP THE TU FOR A FILE-SCOPE TYPEDEF THE DRAFT ALSO CARRIES (P31 S71; byte-proven `ov_SC03_092/func_8017FA74`) L33790 - **§426** — ★★★ — main's SWITCH FUNCTIONS WERE NEVER A CODEGEN WALL: ONE RODATA CARVE HAD BEEN MISSING SINCE PHASE 7 (P31 S72; 3 of the 11 "PROVEN gate-rejects" banked byte-identical in 14 s) L33865 - **§427** — ★★ — A HASH IS A CORRECTNESS ORACLE WITH ZERO DIAGNOSTIC CONTENT; PRESERVE THE RED ARTIFACT BEFORE ANYTHING REBUILDS OVER IT (P31 S72) L33945 - **§436** — ★★★ — TWO TOOLS THAT READ THE WRONG SOURCE OF TRUTH, AND THE SHAPE THEY SHARE (P31 S74; both fixed + negative-controlled) L34374 - **§444** — ★★★ — A REJECTED GATE DAMAGED COMMITTED CARVE STATE, AND `git status src/` SAID NOTHING WAS WRONG (P31 S75) L34739 - **§446** — ★★★ — "RELOC-ONLY REMAP GATED DIFF" IS A CARVE VERDICT, NOT A CODEGEN VERDICT, UNTIL YOU DIFF THE `.text` (P31 S75; 4 siblings + 1 latent, ~1,300 ins) L34817 - **§447** — ★★★ — THE BIGGEST FUNCTION LEFT IS A CARVE, AND THE VERDICT TOOL SAID "DECLARATIONS" (P31 S75; SaveLoadRoutine, 1,165 ins) L34864 - **§449** — ★★★ — FOUR COMPILER DIALS FROM THE S75 REDRAFT WAVE (two byte-gate CONFIRMED, and one correction to §439) L34951 - **ADDENDUM** — to §265 (P31 S76) — THE PROSE GUARD DID NOT HOLD. IT IS A GATE REFUSAL NOW. L35147 - **CORRECTION** — to §182/§188 (P31 S76) — THE EPILOGUE "WALL" IS AN ORACLE ARTIFACT IN THE REORDER ISLAND L35192 - **§478** — 🔴 — A VERBATIM DRAFT IS THE STRONGEST *FALSE* SIGNAL YOUR SCOPING TOOL CAN EMIT L35797 - **§483** — ★★★ — THE S77w WAVE HARVEST: SIX LEVERS, FOUR FROM BANKED (BYTE-PROVEN) BODIES L35977 - **§484** — ★★★ — "NO SINGLE NOLOAD BASE" IS NOT "UNLINKABLE": ASK WHETHER THE `.bss` OFFSETS ARE DISJOINT (P31 S77) L36033 - **§485** — ★★★ — THE PLACEMENT MAP WAS PARSING A PRETTY-PRINTER: 25 PsyQ OBJECTS WERE INVISIBLE, NOT ABSENT (P31 S77) L36079 - **§487** — ★★★ — THE PSX LOADER'S PER-VERSION SIGNATURE SETS ARE A FREE PROVENANCE ORACLE: main's "WALL" BAND IS LIBPAD 4.2.1 (P31 S78) L36160 - **§488** — ★★ — THE "GAME CODE" GAPS BETWEEN LIBRARY BLOCKS WERE LIBRARY OBJECTS: 13 SUBSEGS → LINKED, EXACT-TILED, ZERO TOKENS (P31 S78) L36212 - **§489** — ★★★ — SCATTERED `.bss` IS A PARTITION PROBLEM: SPLIT THE SECTION INTO PER-BASE PIECES AT LINK-PREPARE (P31 S78 #4; closes the §9.1 exclusion class 3/3) L36252 - **§490** — ★★★ — THE "WALL" BAND WAS A LIBRARY VERSION AWAY: LIBPAD 4.2.1 + LIBAPI 4.2 FOUND, THE WHOLE 0x8005CE18–0x8005FC68 BAND + THE APICARD REGION LINKED FROM REAL OBJECTS (P31 S79 #13/#5) L36310 - **§491** — ★★ — THE MECHANICAL LEFTOVERS (P31 S79 #6): A PHANTOM STUB, TWO JTBL TWINS, ONE EXACT CLONE — AND THREE TOOL GAPS THE BANKS EXPOSED L36356 - **§492** — ★★ — "C-PLUMBING" WAS THREE DIFFERENT THINGS (P31 S79 #7): A RAW SPLICE THE GATE'S LADDER BROKE, TWO -O0 BODIES THE CHECKER COMPILED AT -O2, AND TWO DRAFTS THAT BELONGED TO OTHER OVERLAYS L36402 - **§493** — ★★ — THE PERMUTER ROUTE END-TO-END, AND THE THREE PLUMBING STEPS BETWEEN A SCORE-0 WINNER AND THE MAIN GATE (P31 S79 #8) L36430 - **§495** — ★★★ — THE VERBATIM END-STATE (P31 S80 #10): TWO DEF-SIDE DECLARATION WALLS, A "BANK" THAT WAS THE ASSEMBLY, AND A GATE THAT DROPPED A BANK ON EXIT 0 L36630 - **§498** — ★★★ — A CARVE THAT DROPS THE BINARY'S `--pre` CLAUSE, AND A GATE THAT IGNORED THE EXTRACT'S EXIT CODE: HOW A BYTE-CORRECT DRAFT WAS BOOKED "DIFF" (P32 T1a; resident `func_800D128C` BANKED 243 ins after two instrument fixes) L36728 - **§499** — ★★★ — A NEVER-ONBOARDED PAYLOAD PINS ITS OWN BASE STATICALLY, AND THE FIRST BUILD CANNOT (P32 T2a/T2b: the parked five onboarded, 213 → 218 binaries, 0 UNCLAIMED) L36758 - **§500** — ★★★ — THE T3 WAVE HARVEST (P32 T3, 2026-09-05): 31 one-agent-per-function drafters → 20 MATCH / 9 NEAR / 2 FAIL, every closer, two NEW named gcc mechanisms, and the wave-process defects L36799 - **§453** — ★★★ — ONE SOURCE PER UNIQUE FUNCTION: THE INCLUDE-AT-SITE SHARE, ITS GATE, AND THE FOUR DEFECTS THAT SHAPED IT (P35 S94–S96) L37561 - **§455** — Rung G, the guided search: score the object, classify the residual, compose the moves (Phase 36 S101) L37692 ### process, measurement & doctrine (144) - **§8e** — The jtbl ALIGNMENT LAW + the pad-spec filter — multi-table .rodata spans (Phase 29, byte-proven; `.run/probe_jtbl/verdict.md`) L549 - **§3-The** — mechanism: game-code dedup is SOURCE-LEVEL, not an object swap (R-D1, the key lesson) L945 - **Per-binary** — toolchain provenance (R24) L978 - **§17a** — The TOOLKIT at WAVE scale (Phase-18 Step-3b/Step-1 — measured) + the pipeline-integration gotchas L1468 - **Diagnostic** — lesson: a failed in-TU build leaves a STALE `.o` L1677 - **Phase-20** — RESOLUTION — `tools/cast_call_sites.py` BUILT + the cap re-diagnosed (R14, byte-proven) L1724 - **§23** — Giant `func_80153E00` cracked (scalar-data CAST); BUT most giant drafts are STALE+INCOMPLETE, and a diag MUST remove artifacts (Phase 21 — byte-proven + a self-correction) L2055 - **§3-h** — _exact OVER-COUNTS ×134 — verify shareability before crediting a class's "reach-134" (R14) L2200 - **§28c** — The close=0 recovery is NOT fully exhausted (§26 corrected, R14); + the dedup_propagate registry-skip recovery (Phase 22 T2) L2375 - **§31-triage** — R17 applies to CODEGEN residuals, never to a compile ERROR (Phase 26 session 7, Drew asked) L2716 - **§41a** — v3.1: the def-side wall was ~71% TOOL-shaped — the five measured defects + the laws that dissolve them (Phase 25 T6, Fable5, 2026-07-09) L2792 - **§41b** — T7 execution: the object-only probe OVER-counts BANKABLE by two link/rodata classes (Phase 25 T7-M1, 2026-07-10) L2847 - **§41b** — addendum — M4 "reconcile_decls" tier is ALSO a probe over-count: 0/8 mechanical (Phase 25 T7-M4, 2026-07-10) L2896 - **§53** — SWEEP A FAMILY WITH THE TOOL ITS EXEMPLAR NEEDED: the jr/switch carve, and how omitting it manufactured the "families don't template" doctrine (Phase 28 T1, 2026-07-15) L4066 - **§3-The** — meta-lesson (R35, and why this one is expensive) L4116 - **§55** — Core-crack wave levers + the GATE-ORCHESTRATION law (Phase 29 T3, 13-agent ultracode wave, 2026-07-17) L4131 - **§55b** — THE GATE-ORCHESTRATION LAW (3 traps, ~3.5h lost; all recovered, 0 data lost) L4160 - **§57a** — Two NSD corrections + the SURGICAL-ONLY law + the honest broad-sweep yield (Phase 29, 2026-07-18) L4341 - **§58** — match_one MATCH ≠ BANK: the four blind spots + the crack-wave reconcile-before-bank law (Phase 29 crack-wave, 2026-07-18) L4381 - **§60** — Classify the residual, don't rank it: the deterministic residual→class classifier and what it measured about the backlog (Phase 29 Task-13A, 2026-07-21) L4445 - **What** — it measured — the whole open backlog, byte-grounded L4469 - **§3-The** — parallel-probe race this surfaced L4501 - **§61b** — The jtbl gate stage: built, and the ORDERING law it exposed (Phase 29 Task-14 stage 4, 2026-07-21) L4696 - **§62** — The jtbl RECONCILE must also follow the carve: the post-carve draft reconcile (`harvest_verify._jtbl_reconcile`, Phase 29 SESSION-11, 2026-07-22) L4871 - **§63** — UPDATE (Phase 29 SESSION-14) — `fix_header_decl`'s "SAFE" verdict is FLEET-BLIND; it MUST be R22-validated L5038 - **§65** — The stranded-draft recovery: BLAST-RADIUS TIERS, and the per-overlay de-macroize that refutes §20's DEF-conflict wall (`tools/demacroize.py` + `tools/blocker_probe.py`, Phase 29 SESSION-16, 2026-07-24) L5053 - **§65a** — The blast-radius taxonomy (makes §61's law structural instead of remembered) L5068 - **§65d** — Existing-ladder baseline, measured (do this before building a recovery stage) L5129 - **§65e** — Two oracles, and the disagreement is the finding (R34 in practice) L5139 - **§65f** — The de-macroize lever's BOUNDARY, measured: byte-neutral 14 times of 15, and the 15th is why the gate exists L5156 - **§66c** — Before a wave, verify the FUEL exists; an "already attempted" set built from the wrong directory lies in BOTH directions L5255 - **§66d-5** — `residual_class`'s "structural ⇒ permuter CPU is waste" is WRONG for schedule permutations (measured, Phase 29 SESSION-18) L5529 - **§3-The** — honest fix was source-level and cheap L5788 - **§3-The** — residual, and the honest read L6298 - **§83** — The parameterised-repeat law, the spill-area trap, and why a per-case edit cannot move a per-case symptom (Phase 29 SESSION-20, `func_80183814` 5,122 ins, cold-ish start → 36 structural / 99.3%) L6450 - **Scope** — , measured (do not over-generalise — §80) L6556 - **§3-Two** — ladder lessons banked with it L6566 - **§3-Do** — the WHOLE axis in one edit, then R22 once L6605 - **§3-THE** — LAW: all-or-nothing PER FAMILY L6638 - **§88b** — the `slti` literal-position law (extends §78 to comparisons) L6724 - **§88e** — a wrong diagnosis, refuted properly (the model for how to treat an inherited lever) L6746 - **§89a** — MEASURE the write set; do not assert its tier (`tools/blast_radius.py`) L6770 - **§90d** — Do not measure a live wave's drafts (§87 in real time) L6848 - **§98** — `conform_decls` had three defects, and only the third needed R22 to find (Phase 29 SESSION-22, `func_8014CF04`) L7151 - **§106** — Persist the MEASUREMENT, derive the POLICY: a stored route let a stale file out-vote the live table (Phase 29 T54, `residual_class._ROUTE`) L7508 - **§125** — Split the CARVE from the BODY before calling a jr residue a wall — and measure it by SHA from a CLEAN tree (P30 SESSION-28; **this section's first draft was WRONG and the method caught it**) L8244 - **§3-The** — meta-lesson L8295 - **Wave** — economics (measured, for the next batch's sizing) L8991 - **§136b** — A prior wave's "genuine byte-DIFF" verdict is NOT reliable evidence (4 of 4 refuted) L9071 - **§136g** — When the index points at the WRONG lever: two byte-refuted routings (func_801863B4) L9249 - **§136h** — CORRECTION: the zero-crack pool does NOT "refill with cheap work" (my error, byte-measured) L9278 - **§136j** — The failure MIX flips with function size (measured across four bands, one session) L9340 - **Rank** — the lane by measured concentration, not by class count L9519 - **§140** — A METRIC IS NOT A MEASUREMENT UNTIL IT IS REPRODUCIBLE FROM THE COMMITTED TREE (P30 S1e: a phantom regression that gated the session's best lever) L9698 - **§3-The** — measurement (do this before any wave; it is ~20 lines and needs no builds) L9828 - **§3-And** — the report-vs-bytes lesson attached to it L9860 - **§146** — RE-MEASURE A WALL BEFORE YOU RESPECT IT. Both "permanent" giants fell to drafts already on disk. (P30 S6, +50,094 ins) L10015 - **§147** — The three-stratum FRAME LAW, and four "stop searching" verdicts (P30 S42, `func_8017C294`, serial run) L10074 - **§3-C.** — Inner-block declaration does NOT delay slot allocation — BYTE-REFUTED L10106 - **§3-D.** — "Cheap fuel" that was never probed: 0 of 31 templatable L10331 - **§3-Two** — corrections to the record L10383 - **§3-The** — two fallouts, and how to close them (both measured, in order) L10433 - **What** — does NOT work (14 byte-measured probes) L10524 - **§157** — the cheap-tier size cliff, measured (S45 p6) L10725 - **§162** — The cross-jump "CALL veto" is a COUNT law, not a CALL law (BOUNDS §88a; P30 S48, `func_80189540`) L11340 - **§3-The** — law L11587 - **DIAGNOSTIC** — TELL — two faces, one law L11629 - **§163** — S48 WAVES 2-3 HARVEST (P30, 2026-08-11/12): the five that were byte-probed and are actionable L11798 - **§163z** — THE UNVETTED REMAINDER (do not cite as law; each needs a dedupe pass) L11866 - **§164z** — REFUTED CLAIMS: do NOT re-derive these L13674 - **§165z** — REFUTED THIS WAVE: do NOT re-derive L14916 - **§167z** — REFUTED IN WAVES 5/6: do NOT re-derive L16212 - **§168** — THE COUSIN TIER (P30 S49, 2026-08-12): h_seq's exact-hash brittleness, measured — and the similarity map above it L16269 - **§176g** — SIZE A WAVE BY INSTRUCTIONS, NOT BY CARDS (P31 S52 — the adopted doctrine) L16998 - **§176h** — THE BATCH-SUBSTITUTION HAZARD MAP (P31 S52): seven holes, three wrong fixes, one law L17036 - **§3-D.** — The measured cost shape, and what to build next L17114 - **§176j-2** — THE REPAIR PASS, MEASURED (do this instead of resuming) L17190 - **§176-C** — 🔴 WALL REFUTATION: a hard-register pin CANNOT schedule around a call, because of a genuine gcc-2.7.2 bug L17754 - **§176-E** — Two cheap source spellings, both cc1-probed L17806 - **§180** — THE LEFTOVER-DRAFT HARVEST: RE-VERIFY WHAT YOU ALREADY HAVE BEFORE DRAFTING ANYTHING NEW L17858 - **§181** — WHAT A WAVE'S GATE ACTUALLY REJECTS (P31 S53, measured on wave R's 45-draft main pile) L17924 - **§182** — §177's HONEST NEGATIVE: the epilogue lever cracked 4 of 16, and the `800c3` cluster held L17980 - **§183** — THE DECLARATION-RECONCILIATION PLAYBOOK (P31 S53, measured on 20 byte-verified drafts) L18006 - **§185** — EDIT THE SIDE THAT IS CHEAP TO VERIFY, AND CHECK A TU RETYPE AT ITS USE SITES L18105 - **§190** — THREE PRESCRIPTIONS FROM THE SAME HARVEST (weaker evidence than §189, honestly labelled) L18346 - **§194-A** — A zero-byte scheduling fence goes AFTER the defining statement to make that computation emit FIRST in its block — and the barrier predicate is `volatile`-or-colon-less, not the `"memory"` clobber (COMPLEMENTS §165-40; does not refute it) L18995 - **§194-D** — Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication + attribution correction of §165-17, NOT a new argument-register law L19162 - **§195-J** — GTE / PsyQ op CALL-vs-INLINE is a PER-SITE SOURCE FACT, not a TU style — both forms coexist in one TU (measured in 2 TUs), and the tell is the target's own opcodes (`lwc2`/`sqr`/`swc2` vs `jal`), not the sibling. Costs -8 ins on func_8018505C. Corollary: the game's inline `sqr` macro emits TWO hazard nops, the SDK's `Square0` body emits ONE — so the inline form is provably not `Square0` inlined (a second instance of §187's SDK-vs-game GTE nop divergence). L20267 - **§201-E** — §194-J-2 — The `last_mem_set` deletion window is measured in SOURCE/expand order, not in the printed stream; a foreign store moved between the pair in C source is a real lever, and `volatile` is not always the better one L21419 - **§230** — THE ANCHOR PROBE: with `%hi`/`%lo` masked, the surviving `addiu` deltas tell you which assignment was written first **(single observation — not yet cross-confirmed)** (P31 S58) L23175 - **§257** — THE DEAD-END LEDGER (P31 S58b): ELEVEN LEVERS THAT MEASURED NULL OR BACKFIRED L24161 - **§194-B** — / §209 addendum (P31 S58b) — TWO MORE INSTANCES, AND THE BOUND IS NOW REFUTED FOUR WAYS L24230 - **§266** — THE INERT-RIDER LAW: A LEVER IS ONLY CITABLE WHEN ITS SOLO REMOVAL BREAKS THE MATCH (P31 S59b; measured 4-of-8 on this batch) L25090 - **ADD-4** — → §167-37 addendum — THE FIFTH PRECONDITION, NOW MEASURED: DOWNSTREAM FIELD RE-READS ⇒ NAME IT **AND** STORE **INSIDE** THE ARM L25404 - **ADD-9** — → §255 "AND CASE-BODY PLACEMENT" bound / §222-addendum-3 — ON A LARGE SPARSE TREE, BODIES FOLLOW **SOURCE** ORDER (measured by a one-word probe) L25481 - **ADDENDUM** — to §172b-1 — TWO PLACEMENT DIALS THE PROMOTION LAW DOESN'T NAME: NARROW THE COUNTER'S DECLARED TYPE, AND MOVE THE CAST INTO THE LOOP BODY L26110 - **What** — I could not verify L27084 - **ADDENDUM** — to §164-75 — the fold-reassociation law also fires at a variable's INITIALIZER, not only a later expression L27141 - **What** — I could not verify L27658 - **ADDENDUM** — to §153 / §236-5 (func_801A419C, md_SC07_003 — waves di and dl, corroborating; refutes a contradicted dj-wave card) L27875 - **What** — I could not verify L28115 - **ADDENDUM** — to §174 Law 4 (func_8017E9A8, ov_SC06_015) L28135 - **ADDENDUM** — the zero-emission-asm family gains a REF-SLIDER PLACEMENT LAW and a paired RESTORER (func_800CB900, md_MAIN_026) L28175 - **What** — I could not verify L28673 - **What** — I could not verify L28979 - **§293** — THE "WHOLE-OBJECT GATE NEEDS EVERY SIBLING" LAW, WRITTEN PROPERLY: IT IS THREE LAWS, AND THE LOAD-BEARING ONE IS "THE BASELINE, NOT THE SIBLINGS" (P31 S61; byte-proven on 15 binaries in one night) L29119 - **§294** — ADDENDA HARVESTED FROM THE S61 GEN0/MAXTOK/MAIN/DEEPSEEK BATCH (waves ab8/ab16/ab24/ab32 · g0a–g0f · m0a · ds1/ds2): 99 candidates, 52 covered, 15 notes → 10 addenda, 29 notes → 5 new laws (§295–§299), 3 refuted L29169 - **§313** — A RECOVERY STAGE THAT REWRITES THE DRAFT IS PART OF THE MEASUREMENT: GATE THE RAW DRAFT FIRST, THE REWRITE ONLY AS A FALLBACK (P31 S65 t5r-recovery; byte-proven func_8017BEBC, 246 ins) L30195 - **REFUTED** — claim — REFUTED — the `j`-slot store is an ASPSX macro-expansion hop, not a cse split; maspsx hard (func_8005DBD8) L30599 - **§NNN** — REFUTED: A `sw %lo(SYM)($at)` IN A JUMP'S DELAY SLOT WITH ITS `lui $at` STRANDED ABOVE THE JUMP IS **ASSEMBLER** MACRO-EXPANSION, NOT A cse SPLIT — OUR PIPELINE CANNOT EMIT IT FROM ANY C, AND THE +1 IS A `maspsx` GAP (P31 S66 round2, `func_8005DBD8`, byte-probed) L30601 - **REFUTED** — claim — REFUTED: a block shared across TWO switch statements is ordinary forward cross_jump, not a (func_8001B0D4) L30648 - **§NNN** — REFUTED: A CODE BLOCK SHARED BY TWO *DIFFERENT* SWITCH STATEMENTS IS ORDINARY FORWARD `cross_jump`, NOT A SHAPE C CANNOT REACH — §164-11 ALREADY BANKS IT FROM PLAIN NESTED SWITCHES (P31 S66 round3, `func_8001B0D4`; ⚠ **UNPROVEN** — gate-REFUSED draft, closeness 68→45, never MATCHed) L30650 - **Addendum** — §194-A addendum — the bare/colon-less fence measured NULL and only the "memory"-clobber fo (func_8017E464) L30771 - **§322** — A PROBE THAT ANSWERS A *NECESSARY BUT NOT SUFFICIENT* QUESTION WILL PRICE BLOCKED WORK AS FREE: RUN THE REAL PLANNER WHEN THE PLANNER IS PURE (P31 S67; measured, 96 of 159 open jtbl functions) L30876 - **§323a** — R53, TWICE IN ONE HOUR: A FAILED BUILD LEAVES THE PREVIOUS BINARY, AND `sha1sum` READS GREEN (P31 S67) L30936 - **§324** — THE ARITY-SELF WALL NEEDS *BOTH* HALVES: A K&R DEFINITION **AND** A NO-PROTO TU DECL; EITHER ALONE STILL FAILS (P31 S67; rtu-proven ov_SC01_005/func_8017FBCC, 14 instances measured) L30959 - **§328** — THE VOLATILE ALIAS MUST BE AN *OBJECT*, NOT A CAST: `*(volatile s32*)&sym` UNFOLDS %lo INTO A SEPARATE `addiu` (P31 S67; byte-proven ov_SC07_007/func_80181B8C, 156 ins, NEW LAW) L31034 - **§332a** — MAIN'S RESIDUAL FRONTIER IS CONTAMINATED WITH TOOLCHAIN WALLS: 4 OF 7 IN ONE WAVE (P31 S67, measured) L31106 - **§337** — THE CC1-ONLY BLOCKER CLASS: `blocker_probe`'s STATIC ORACLE REPORTS "none" AND THE WHOLE-TU COMPILE STILL FAILS (P31 S67; ov_SC04_011/func_801827DC, md_MAIN_027/func_800CB4A4) L31167 - **§342** — A `void *` PARAMETER CAST TO ITS REAL TYPE IN A LOCAL IS **NOT** BYTE-NEUTRAL WHEN A LATER PARAMETER ALSO NEEDS A CALLEE-SAVED REGISTER (P31 S67; byte-proven ov_SC07_002/func_80181394, NEW LAW) L31267 - **§343** — `decl_prior`'s FLEET MAJORITY CAN BE WRONG ABOUT THE TRUE SIGNATURE — READ THE RIVALS, NOT JUST THE WINNER (P31 S67; measured on func_8012BD14 / func_8012D624 / func_80143C74) L31284 - **§352** — ⚠ TWO IDENTICAL `__asm__ __volatile__("")` BARRIERS **MERGE WITH EACH OTHER** — SPELL THE SECOND ONE DIFFERENTLY (P31 S67; byte-proven resident/func_800D128C, measured closeness 105 when they merged) L31496 - **§356** — MEASURE A DRAFT IN THE TU IT WILL LIVE IN, NOT IN THE STANDALONE PROBE (P31 S68; measured over 47 stored drafts) L31584 - **§361** — ★ — A LOOP-TAIL BYTE SIGNATURE THAT NAMES ITS SOURCE SHAPE — AND THE LAW THAT A "SCHEDULING TIE" MAY BE YOUR OWN EARLIER LEVER (P31 S68; byte-proven main/func_800241C0, fable escalation, 19 → 0 in 3 iterations) L31647 - **§370** — ★★ — A **HARD BOUND** FROM sched.c, AND THE reorg SLOT-STEAL DIAGNOSTIC (P31 S68; main/func_8001BC6C, 69 ins, NOT closed — 33 → 28 over ~45 measured compiles) L31848 - **§376** — ★★★ — A STANDALONE `match_one` CLOSENESS OF 0 IS A CLAIM ABOUT THE **BODY**, NEVER ABOUT THE **TU** (P31 S69; measured 0/28) L32065 - **§384** — ★★★ — A CARVE-CONFIG BANK IS RED UNTIL YOU RE-EXTRACT, AND THAT LOOKS EXACTLY LIKE A FALSE BANK (P31 S69; measured twice, cost one destroyed match) L32269 - **§378b** — ★★★ — THE FOUR VARIANTS OF THE DECL BLOCKER, AND THE TWO PLACES §378 DOES **NOT** APPLY (P31 S69, all four measured the same day) L32582 - **§398** — ★★★ — `family_remap` CARRIES THE **SOURCE** TU's DECL ENVIRONMENT INTO A DESTINATION THAT ALREADY OWNS THOSE NAMES (P31 S69; measured 3 banked of 22) L32845 - **§3-E.** — WHAT THE AGENTS REFUTED L33158 - **§408** — ★★★ — §406 REFUTED AS A SWEEP: THE SHAPE IS THE FAMILY, THE DISAGREEMENT IS THE DEFECT (P31 S71; 0 MATCH / 14 applied, 0 / 210) L33219 - **§411** — ★★★ — THE PACK MUST CARRY THAT FUNCTION'S OWN HISTORY (P31 S71; measured 38/39 vs 124/131) L33396 - **§428a** — ★★★ — TWO RESIDUALS THAT MOVE IN OPPOSITE DIRECTIONS UNDER EVERY LEVER USUALLY SHARE ONE CAUSE (P31 S72; `main/func_8001B0D4`, NEAR/53 -> MATCH; **my first answer here was WRONG and is kept below as the refutation**) L34002 - **§430** — ★★★ — A GOTO INTO A LOOP IS FINE; HAND-HOIST THE CONSTANTS IT COSTS YOU (P31 S73 — **this section previously said the OPPOSITE and was wrong; the refutation is kept below**) L34083 - **§439** — ★★ — LEVER SET FROM THE S74 WAVE (each entry is one measured crack, not a hypothesis) L34565 - **§449** — ★★★ — FOUR COMPILER DIALS FROM THE S75 REDRAFT WAVE (two byte-gate CONFIRMED, and one correction to §439) L34951 - **§3-C.** — CORRECTION TO §439 — `sll 16; srl 16` LANDS *AFTER* THE CALL, AND WORKS FOR A KNOWN CONSTANT L35000 - **§469** — THE `MEM_IN_STRUCT_P` ALIAS UNLOCK (and §463's spill law, independently confirmed) L35424 - **§473** — 🔴 §265's "HANDWRITTEN" VERDICT FOR `ov_SC07_002:func_8017DC80` IS REFUTED (324 → 89) L35529 - **§463** — 🔴 SPILL SLOTS ARE 8 BYTES, AND THE §41b "LOAD ABOVE THE PROLOGUE" WALL IS REFUTED L35687 - **§479** — ★★★ — WHERE THE PERMUTER ACTUALLY PAYS: A MEASURED YIELD CURVE (P31 S77, 8 candidates) L35817 - **§501** — ★★★ — A LEVER THAT MEASURES WORSE MAY BE A CASCADE: READ THE `.loop` DUMP FOR THE DESIRABILITY FLIP BEFORE DISCARDING IT (P32 T4b, `main:func_800391D4`, a pinned wall banked by a Fable agent) L37157 - **§454a** — the recipe rung's real shape: replication, not discovery (S99, measured) L37677 - **§458** — Struct spelling is a per-access dial: the T2 probe's byte evidence (Phase 37 S106) L37886 ### (unbucketed — title matched no symptom vocabulary) (338) - **§3-How** — to use this L30 - **§1** — Idiom catalog (asm pattern → C that produces it) L39 - **§3-I3** — Division by a constant → magic multiply L54 - **§3-I4** — Runtime (variable) division → `divu` + zero-check `break` (NEEDS `--expand-div`) L60 - **§2** — Writing matching C (what makes gcc emit X) L69 - **§3a** — Escalation TIER above the permuter — web-research the compiler internals (HIGH VALUE, proven) L128 - **§3b** — §31-directed permuter mutation — bias the search over the class's levers (Phase 24 T5) L147 - **Residual** — B — a `return ` materialised late / merged instead of distributed per-site L886 - **§3-The** — A→B→C per-overlay workflow L1110 - **Dedup-credit** — (§11) for overlays — two shapes L1127 - **§14a** — The fleet bulk run (Phase 15 — `--auto-from`, 947 REAL → 74,527; 3.82% → 22.14% in one pass) L1192 - **§14d** — Deterministic recovery beats agent waves on the hard tail (Phase 15, the final-session finding) L1292 - **§14e** — Two hard-tail dead-ends (Phase 15 — documented so they aren't re-attempted) L1321 - **§3-The** — STEERABLE idioms (byte-confirmed this phase) L1427 - **Strategic** — conclusion — the match-% lever post-research L1458 - **§3-NEW** — / CONFIRMED residual classes (this session) L1683 - **Operational** — gotchas (cost real time) L1704 - **What** — WORKED — the Phase-20 reusable wins L1712 - **§21** — wave-distilled idioms (Phase 21) L1775 - **§26** — The cheap close=0 recovery lever is EXHAUSTED; `idiom_loop --assess` was DOUBLY inflated (Phase 21, cont.7) L2222 - **§3-The** — `_a` close=0 recovery banks 0/20 — same def-side wall as MAIN (0/40) L2228 - **§27** — Giant matching recipe (Phase 21 cont.7b — validated on func_80176D94, 152 ins) L2282 - **§28a** — decomp.wiki GCC patterns worth trying on BFM giants (decomp.wiki/compilers/GCC, raw at decompals/decompedia; PS1-applicable subset) L2368 - **§31** — THE gcc-2.7.2 CODEGEN MAP: pass → residual → C-lever catalog (Phase 23; 4 Fable5 agents read the compiler source) L2425 - **§36** — Fable5 giant-crack levers (Phase 24 T7 Fable5 batch; accumulates as each lands — full RTL dumps in `.run/t7/fable/`) L2510 - **§42d** — addendum — wave 4 (rtu_match fan-out over the mapped frontier, 2026-07-10c): 24/26 MATCH, +5 durable levers L3120 - **§3-B.** — THE EBB RULE — the general form of §46-L2 L3506 - **§3-E.** — Meta L3556 - **§50** — Refinements that BOUND §47/§48 (from the `func_80135EB0` wall, 21→6; Phase 26 session 8) L3614 - **§51a** — The bug class L3674 - **§51c** — THE METHOD (do not audit by reading the regex) L3700 - **§51d** — THE LAWS L3715 - **§51e** — The false-wall pipeline (why this is not just hygiene) L3777 - **§3-Why** — the wall is (probably) intrinsic L3983 - **§3-The** — case L4071 - **§3-The** — rule L4100 - **§55a** — New byte-proven levers (each from a banked or near draft) L4136 - **§60a** — What the first DIRECTED grinder run exposed (Phase 29 Task-13B, 2026-07-21) L4510 - **§60b** — The plateau autopsy's verdict: a `partial` drift is a WRONG DRAFT, not a missing transform (Phase 29 Task-13B close, 2026-07-21) L4548 - **§66** — Exercise a banking driver's SUCCESS path before pointing it at a wave: the free re-bank test (Phase 29 SESSION-17, 2026-07-24) L5197 - **§66d-2** — Two operational sharp edges L5327 - **§66d-3** — Read the ILS per-cycle SERIES, not its final best: a repeated score and a still-falling one look identical in a summary line and mean opposite things L5338 - **§67** — The arg-copy PLACEMENT lever: launder a parameter into a fresh pseudo AT the statement where the target's copy lands (Phase 29 SESSION-18, `func_8014D820` 25 → 16) L5353 - **§66d-4** — "ILS converged" means converged FOR THAT WEIGHT PROFILE, not a floor (amends §66d-3; Phase 29 SESSION-18) L5476 - **§69** — How to attack a behemoth: map it, don't draft it (Phase 29 SESSION-18, `func_80183814`, 5,122 ins) L5607 - **What** — is left, and what is byte-recorded as SPENT L5804 - **§3-The** — mechanism, with citations L6075 - **§3-Two** — diagnosis traps this function proved L6102 - **Practice** — Practice L6112 - **§3-The** — drift was an allocation decision, not missing code L6218 - **§3-The** — economics L6249 - **§71** — has a blind spot, and this is it L6264 - **§3-The** — flagged "#1 move" LOST — and why the failure is informative L6340 - **§78** — 's attribution primitive, run and reproduced L6350 - **Cold-start** — economics, now complete L6356 - **Also** — reproduced on this function L6439 - **§3-And** — the banking footnote (§75a class A, one line) L6443 - **§83a** — READ THE HEADLINE NUMBER CORRECTLY: a LENGTH drift makes `match_one`'s count meaningless L6457 - **§83b** — THE LEVER: find the parameterised REPEAT before decoding case-by-case L6465 - **§83e** — §80 vindicated again, on the same day it was written L6504 - **§3-The** — conflict L6582 - **§3-THE** — FAILURE MODE — widening only `engine_core.h` is worse than not starting L6590 - **§3-The** — precondition, and how to check it in one grep L6597 - **Reading** — , for the next person L6614 - **§3-The** — guard refuses a class that largely works L6629 - **§3-The** — blindness ladder, now complete — FOUR classes `match_one` cannot see L6687 - **Consequence** — for the backlog ledger L6697 - **§88** — `cross_jump` will not merge a common suffix containing a CALL; and §78 is scoped to ORDERED comparisons only (Phase 29 SESSION-20, the behemoth close-out) L6712 - **§88a** — repeated CALL-shaped blocks are left UNMERGED; call-free tails are merged for you L6717 - **§89** — Two throughput rules the project already had written down and was not following (Phase 29 SESSION-20) L6764 - **§3-The** — standing sequence L6796 - **§90b** — "Byte-neutral" is not "wanted": undo on the SUCCESS path too L6827 - **§90c** — A library-callable function must FAIL CLOSED on an unconfigured module L6837 - **§90e** — An agent's CONCLUSION and its EVIDENCE fail independently — re-derive the premise, not the fix L6856 - **§93** — `set -o pipefail` attributes a pipeline failure to the LAST stage, not the failing one (Phase 29 SESSION-21, `func_8014D820`) L6960 - **§101** — The STALE DEFAULT class: a guard whose cause was removed is a silent skip wearing a safety label (Phase 29 SESSION-22, three instances in one session) L7283 - **§110** — A unit must define exactly ONE function, and "ends in `;`" does not tell you which line defines it (Phase 29 T65) L7678 - **§115** — A `func_XXXXXXXX` predicate rots by design: the same name-form assumption in THREE places (Phase 29 T78) L7888 - **§3-The** — fix moves the DEFINITION, not the stub — and here is why the obvious shortcut fails L7933 - **§117** — Spell the sibling's symbol from the SIBLING's address, not the exemplar's kind (Phase 29 T82) L7964 - **§118** — Ordinal (positional) immediate resolution: compare C tokens to the DIFFERING asm uses (Phase 29 T87) L8001 - **§119** — Two levers on the SAME axis, opposite directions: test the off-diagonal (Phase 29 T89) L8035 - **§3-The** — wiring trap that cost two attempts L8081 - **§3-The** — method (keep this) L8251 - **§3-The** — instrument rules that make its answer trustworthy (this is where I failed) L8261 - **§3-The** — corrected results (each SHA-verified, from a clean tree, restore re-verified) L8276 - **§3-The** — instrument trap that hid it (and it is §124's shape again) L8328 - **§3-The** — mechanics L8337 - **§3-The** — idiom they kept re-deriving: the constant-offset fold L8394 - **§127b** — the knowledge was in a SOURCE COMMENT, not the cookbook L8424 - **§128** — A raw NUL in C source makes grep SILENTLY SKIP the file (P30 S28, 137 files) L8431 - **§128a** — a negative control must corrupt a SCRATCH COPY, never the tracked file L8463 - **§3-The** — real blocker underneath, for the record L8515 - **§3-The** — diagnostic ladder that finally located it (reusable) L8561 - **Defect** — 1 — a pre-§8e MERGED DOUBLE is not a single-table predecessor L8617 - **Defect** — 2 — `as` writes a corpse and nothing deletes it L8635 - **§3-The** — fingerprint, and the 30-second ladder that found it L8645 - **§3-The** — transferable rule L8666 - **§132a** — `--like` is for a sibling with NO record; against one that HAS a record it over-derives (P30 S29, `ov_SC07_010`) L8673 - **§135** — Six byte-verified gcc-2.7.2 idioms from the P30 S6f-h waves (and the two-lane wave shape that found them) L8803 - **§3-The** — codegen idioms L8808 - **§3-The** — wave shape that produced these L8857 - **§136** — The LOCAL-VARIABLE lever: how many C locals, at what scope (P30 wave 4a, 25 byte-verified banks) L8875 - **§136a** — Blocker capture: classify on the OUTPUT, never on the exit status L9014 - **§136d** — Four gcc-2.7.2 levers the redraft lane found (each closed a residual no other lever moved) L9132 - **§136e** — §136c's PRECONDITION, and two more symptom keys (wave 4b batch 3) L9176 - **§136i** — The drafter model LADDER: Haiku → **Sonnet** → Opus → Fable5 (Drew, 2026-08-03) L9308 - **§3-The** — triage, cheapest first L9459 - **THREE** — carry variants hide in one "CARRY-FIXABLE" bucket — and they need different fixes L9528 - **STEP** — 0 of sibling-first: grep `src/` for a distinctive LITERAL from the `.s` L9584 - **§3-The** — generalisation — three corollaries worth more than the bug L9664 - **§3-And** — the inverse-lookup trap, same session L9681 - **§3-The** — three-line proof (do this before diagnosing any metric movement) L9718 - **§3-The** — two instrument defects it exposed L9727 - **§3-Two** — wrong mechanisms I chased first, and why they were wrong L9758 - **§3-The** — trap that hid it — SAME FUNCTION, TWO ROUTES, ONLY ONE IS FREE L9843 - **Route** — selection (why `--addr` sometimes says "nothing changed") L9852 - **§144** — THE LITERAL'S SPELLING PICKS THE IMMEDIATE ENCODING (P30 S40 wave 1, `func_801822E0`) L9926 - **§3-Why** — a correct draft can read as an intrinsic wall L10026 - **§3-The** — rule L10061 - **§3-B.** — `(v < 0x40) ? v : 0x3F` is folded to `MIN_EXPR` and expands to the WRONG SHAPE L10208 - **§3-D.** — Reproduce the original's BUGS verbatim L10232 - **§149** — Four instrument defects in one session, and the two questions they were hiding (P30 S43) L10281 - **§3-A.** — A prep step that returns its input on failure is indistinguishable from a search that found nothing L10287 - **§3-B.** — Same address + same name ≠ same body — and the ledger keys on address L10305 - **§3-C.** — `make: *** [...] Error N` is a summary, never a diagnosis L10321 - **§3-The** — fix L10357 - **§3-The** — method that found it (this is the transferable part) L10367 - **Diagnostic** — order (adopt this) L10394 - **§3-The** — mechanism (read from cc1's own `-dR` trace, not inferred) L10410 - **§3-The** — lever — a zero-emission insn that absorbs the blocked tick L10425 - **§3-The** — finding L10461 - **§3-The** — key L10470 - **§3-Two** — cautions that must travel with this technique L10481 - **§3-The** — companion defect (open) L10492 - **Symptom** — Symptom L10512 - **Mechanism** — (gcc source + RTL dumps, not inferred) L10517 - **§3-The** — cure — a fresh launder per site, each in its own block L10530 - **Companion** — levers from the same function L10540 - **§3-B.** — Two static base-derivation methods that must AGREE (use both) L10574 - **§155a** — the same failure class, one level up: SHAPE-blind table scanning (S45 p5) L10605 - **§155c** — the ZERO-REFERENCE trap: gcc splits a global-array address across the `lui` and the LOAD (S46) L10648 - **Symptom** — Symptom L10759 - **§3-Why** — the fork is chained (gcc source, validated insn-by-insn against -dS/-dR dumps) L10765 - **§3-The** — method (dump-arithmetic first, then place — no probing) L10778 - **§162a** — SHARPENS *(sharpens §161a, §131, §8a-pad, §129a)* L11079 - **§162b** — SHARPENS *(sharpens §48-A3, §156, §150, §76)* L11108 - **§162d** — SHARPENS *(sharpens §31, §21, §30, §55a)* L11137 - **§162e** — NEW L11179 - **§162f** — SHARPENS *(sharpens §42d, §41d, §73, §10)* L11247 - **§162g** — NEW L11302 - **§162h** — SHARPENS *(sharpens §88, §88a, §50-B, §8)* L11338 - **§162i** — SHARPENS *(sharpens §135, §21, §42, §32)* L11406 - **§162j** — SHARPENS *(sharpens §25, §136d-1, §48-B, §46-L2)* L11431 - **§162k** — SHARPENS *(sharpens §1-I2, §12, §160d, §21)* L11460 - **§162l** — SHARPENS *(sharpens §48-B, §48-C1, §20, §21)* L11523 - **§162m** — SHARPENS *(sharpens §36, §158, §148, §153)* L11578 - **§158a** — THE FIFTH LEVER IS NOT AN ASM: `do { } while (0)` is a REGION ref-multiplier you MINT (P30 S48, `func_8017CBC8`, ov_MAIN_012 / jr_801789AC, 188 ins → MATCH) L11580 - **Size** — it before you write it (§158 step 1-2, applied) L11598 - **§3-The** — wrap BOUNDARY is the dial — and it is indiscriminate L11619 - **§3-Not** — a pure dial L11625 - **§162n** — NEW L11651 - **§162o** — SHARPENS *(sharpens §158, §136-1, §136-6, §79)* L11684 - **§162p** — SHARPENS *(sharpens §48-B, §46-L2, §156, §136d-1)* L11743 - **§162q** — SHARPENS *(sharpens §30, §30a, §135-2, §136-13)* L11760 - **§164** — S48 §163z SKEPTIC PASS (P30, 2026-08-12): 190 claims vetted, 82 banked L11886 - **§169** — THE MICRO-ADAPT LANE (P30 S49): edit a proven body, don't crack a new one L16321 - **§170** — THE A-PROP WORD-DIFF CARD (P30 S49): the lane that had no owner L16368 - **§171** — THE STALE SEED SYMBOL (P30 S50, 2026-08-13): why §170's 91%→57% was never codegen L16412 - **§172** — THE ORPHAN-SLOT MECHANISM v2 (P30 S50-Max): the complete frame-residue model for gcc-2.7.2 MIPS L16561 - **§172a** — TWO DECOMPILATION TELLS FROM THE SAME DIG (P30 S50-Max) L16615 - **§172b** — THREE MORE TELLS FROM THE GCC READ (P30 S50-Max, banked on Drew's ask) L16631 - **§174** — THE ADAPT-CARD WAVE RECIPE (P31 waves A/B, 2026-08-14): prevention beats recovery L16710 - **§176a** — THE VERIFICATION-LAYER LAWS (P31 overnight, 2026-08-15). What each check can and cannot prove. L16805 - **§176b** — BATCH-GATING MECHANICS (P31): what changes when N drafts land in ONE .c L16845 - **§3-C.** — The limit that remains (recorded, not solved) L17084 - **§176k** — TWO SELECTOR BUGS THAT SILENTLY SHRINK A WAVE L17210 - **§178** — SIX LEVERS MINED FROM THE WAVE-P JOURNALS (P31 S52), each byte-proven and source-cited L17284 - **§3-A.** — THE `$0`-ADD OPAQUE COPY defeats `make_regs_eqv` (func_80033398, 93 ins) L17296 - **§3-C.** — SINGLE-SET TEMPS GET THE BIRTHING BOOST (func_8001D3FC, 196 ins) L17325 - **§3-E.** — THE ZERO-OFFSET ALIAS HOLE (func_80037028, 71 ins) L17340 - **§179-B** — 🔴 THE HAND-WRITTEN-ASM TRANSCRIPTION CHECKLIST (maspsx + masked_diff; 10 byte-proofs) L17430 - **§179-C** — 🔴 A FUNCTION WITH NO EPILOGUE (falls into a sibling's shared tail) MUST BE FILE-SCOPE `__asm__` L17480 - **§179-D** — `gte_stflg` MUST CLOBBER `"$12"` OR THE WHOLE TEMP FILE ROTATES BY ONE L17531 - **§179-H** — A MID-BODY `.global LABEL` PAIR SLICES A BYTE-COMPARABLE FRAGMENT OUT OF A LARGER ROUTINE L17627 - **Considered** — and NOT banked L17650 - **§176-F** — Misdiagnosis triage: four residual verdicts that were lying L17828 - **What** — is NOT banked here L17845 - **§180c** — WHEN A BINARY'S MASS BAND IS SPENT, THE FLEET-WIDE DRAW IS STRICTLY BETTER L17906 - **Only** — ONE of 27 blocked drafts was wrong. The other 26 were correct and unbankable. L17925 - **§180d** — THE `pgrep` BRACKET TRICK PROTECTS THE PATTERN, NOT THE COMMAND LINE L17994 - **Three** — defects in one call path; the overlay slates that carry most of the wave work were being waved through L18406 - **§194-F** — `if ((*p = v = f()) == 0)` is an expand-time pseudo SPLITTER (store_expr's `want_value && MEM` path), not a fold — it partitions one value between local_alloc and global_alloc. BOUNDS §21's L1872 bullet, whose stated direction is byte-wrong on 3 of 4 instances, and CLOSES the control §167-37 asked for. L19240 - **§194-L** — §88b and §189-E are BOTH half-wrong, but not the way the candidate says: the compare-constant shape is a 2-D lookup (cmp_info ROW × constant-in-window), and naming matters on OPPOSITE sides of the window for the GE/LT rows vs the GT/LE rows L19567 - **§197-REJECTED** — §197-REJECTED L20743 - **§199-REJECTED** — §199-REJECTED L21139 - **§201-B** — In a narrowed PLUS/MINUS/AND/IOR/XOR expression the destination pointee is INERT — the sign of the materialized constant is decided by an OR over the UNWIDENED operands (convert.c trunc1), which bounds §1841 to direct constant stores L21241 - **§201-REJECTED** — eight, the session's highest L21475 - **§204-CONFIRMED** — 30 reports that the index already answered L21941 - **§204-REJECTED** — sixteen, twice the previous record L22035 - **§209** — THE NARROW LOCAL IS A DIAL IN TWO OPPOSITE DIRECTIONS, AND §194-B's "≥2 `sh` STORES" BOUND IS BYTE-WRONG (P31 S58) L22302 - **§212** — THE WALKING CURSOR IS COUNTABLE: `*wp++` emits one `addiu` PER STORE, `wp[0..2]` emits one (P31 S58) L22484 - **§213** — INDEPENDENT SAME-BASE STORES: THE EMISSION ORDER IS A PERMUTATION OF SOURCE ORDER, AND THE PERMUTATION IS NOT ALWAYS THE IDENTITY (P31 S58) L22529 - **§217** — DECODING A CALL'S STACK ARGUMENT SLOTS: `sw` at `0x10`/`0x14`/`0x18` are params 5/6/7 **(single observation — not yet cross-confirmed)** (P31 S58) L22707 - **§226** — FRAME PADS: FOUR WAYS §162i1/§2429's DEAD-LOCAL LEVER MISFIRES (P31 S58) L23028 - **§228** — READING THE DIVIDE, PART N: the off-by-one compare is `% K == 1`, and three more discriminators (P31 S58) L23089 - **§229** — THE ADDRESS IS A VALUE: NAMING IT MOVES THE `lui`/`addiu` PAIR — AND §L14410 SAYS THE OPPOSITE FOR A REASON (P31 S58) L23126 - **§231** — TRANSCRIPTION AND SEMANTIC-READ HYGIENE: six ways the listing misleads (P31 S58) L23190 - **§194-B** — addendum (P31 S58) — BOUND 2 is byte-wrong; see §209 Direction A L23287 - **§176-B** — addendum (P31 S58) — the misdiagnosis direction L23293 - **§164-63** — addendum (P31 S58) — the interposed asm works when the SECOND value is an ordinary assignment L23308 - **§193-A** — / §194-E addendum (P31 S58) — where the twin's body actually lives L23317 - **§235** — THE PHANTOM SYMBOL: A MASKED `MATCH` CAN CARRY A RELOCATION THAT DOES NOT EXIST (P31 S58b) L23408 - **§238** — SAME NAME, DIFFERENT FUNCTION: THE OVERLAY-HOMONYM TRAP (P31 S58b) L23571 - **§240** — `A + K + B`: WRITE THE CONSTANT **BETWEEN** THE TWO RUNTIME TERMS (P31 S58b) L23645 - **§241** — THE FOLDED SIGN-EXTEND-AND-SCALE: `sll 16 ; sra (16 − log2 scale)` (P31 S58b) L23674 - **§252** — THE GUARDED PRE-DECREMENT: `(x != 0) && (--x == 0)` (P31 S58b) L24042 - **§255** — THE EMPTY CASE, PART 2: FOUR TREE SHAPES IT BUYS (P31 S58b) L24089 - **§258** — ADDENDA TO EXISTING SECTIONS (P31 S58b) L24204 - **§202** — addendum (P31 S58b) — THE DEF-SIDE ALIAS ALSO CLEARS A RETURN+PARAM DOUBLE CONFLICT L24253 - **§205** — addendum (P31 S58b) — CHAINED ASSIGNMENT: N≥3 IS INNERMOST-FIRST, AND THE TEXT MIRRORS EMISSION L24264 - **§208** — addendum (P31 S58b) — IT SCALES TO SIX SITES, AND IT HAS AN EXACT INVERSE L24281 - **§210** — addendum (P31 S58b) — THREE CONFIRMED SPELLINGS OF THE BOOLEAN TAIL L24310 - **§213** — addendum (P31 S58b) — THREE MORE PERMUTATION LAWS FOR INDEPENDENT SAME-BASE STORES L24369 - **§217** — CROSS-CONFIRMED (P31 S58b) — AND THE INCOMING HOME SLOT IS THE MIRROR L24486 - **§220** — addendum (P31 S58b) — THE PARAMETER, NOT A COPY (SEVEN CARDS) L24503 - **§225** — addendum (P31 S58b) — THE GUARD-CLAUSE FINGERPRINT, AND THREE MORE SHAPES L24600 - **§229** — addendum (P31 S58b) — NAME IT **INSIDE** THE ARM L24672 - **§230** — CROSS-CONFIRMED (P31 S58b) L24683 - **§231** — addendum (P31 S58b) — FOUR MORE WAYS THE LISTING MISLEADS L24692 - **§232** — CROSS-CONFIRMED (P31 S58b) L24723 - **§260** — THE §154-A LEADING-ISLAND SPLIT: ONE CONFIG LINE, AND THE ISLAND PEELS FROM THE END (P31 S59, byte-proven) L24773 - **§262** — A LANE'S YIELD IS ONLY A LANE FACT IF IT IS SIZE-MATCHED (P31 S59) L24907 - **§264** — FOUR TELLS-LANE C RECIPES, EACH DRIVEN TO MATCH (P31 S59) L24978 - **ADD-1** — → §231 addendum (also cross-ref from §195-D) — THE MASKED-`jal` "MISSING CALL" ILLUSION L25138 - **ADD-6** — → §172b-1 / §264 addendum — SHIFT-AS-TEST: `(x << 16) != 0` TESTS THE LOW HALF WITHOUT TRUNCATING THE PSEUDO L25192 - **ADD-7** — → §256 addendum — THE SINGLE-GUARD GOTO: THEN-BLOCK OUT OF LINE AT THE TAIL L25201 - **ADD-3** — → §238 addendum — TWO BINARIES CAN EACH DEFINE THE SAME `func_` NAME WITH DIFFERENT BYTES, AND NO GENERATED REPORT SHOWS IT L25391 - **§270** — The A-prop 0-bank anatomy: a byte-correct body still needs FOUR layers to agree (P31 S59) L25569 - **ADDENDUM** — to §1-I5 L25859 - **ADDENDUM** — to §164-51 L25924 - **ADDENDUM** — to §176-F5 L25940 - **ADDENDUM** — to §225 L25969 - **ADDENDUM** — to §229 — A POINTER'S NUMBER OF USES DECIDES WHETHER ITS ADDRESS FOLDS OR SURVIVES A CALL L26157 - **ADDENDUM** — to §226 — A DEAD LOCAL SCOPED TO A NESTED BLOCK BUYS A MID-FUNCTION `addiu sp` PAIR, NOT A FRAME-SIZE CHANGE L26275 - **ADDENDUM** — to §74 (func_800D0E30, resident) L26643 - **ADDENDUM** — to §172b-4 (func_80185054, ov_SC03_097) L26679 - **ADDENDUM** — to §265 (func_8017DC80, ov_SC07_002) L26717 - **ADDENDUM** — to §17 (func_800CB794, md_MAIN_036) L26756 - **ADDENDUM** — to §162p (func_8017C120, ov_MAIN_012) L26804 - **ADDENDUM** — to §20 (~L1947) (func_80186AD0, ov_SC06_032) L26838 - **ADDENDUM** — to §164-56 (func_8017F644, ov_SC04_005) L26868 - **ADDENDUM** — to §237 (func_8017F7FC, ov_SC03_092) L26903 - **ADDENDUM** — to §195-E — a goto-ladder's STORES must sit AT the labels, after the gotos, not inline before them L27255 - **Harness-defect** — flags L27729 - **ADDENDUM** — to §167-40 (func_8017E2CC, ov_SC04_015 — wave dg) L27822 - **ADDENDUM** — to §87 (func_801815F4 ov_SC06_032; corroborating func_801840DC ov_SC05_017, func_80189C68 ov_SC03_006 — wave dg) L27856 - **ADDENDUM** — to §263 (func_801E83AC, md_SC04_029 — wave dj) L27895 - **ADDENDUM** — to §265 (func_8017D878, ov_SC03_107 — wave dj; corroborated by a REJECTED, contradicted card in wave dm — see closing) L27909 - **ADDENDUM** — to §6 (func_801811F0, ov_SC03_102 — waves dj and dl, corroborated by a self-reported "nothing new" dm-wave card) L27923 - **ADDENDUM** — to §195-G (func_80183BB0, ov_SC05_001 — wave dj) L27937 - **ADDENDUM** — to §74 (func_8017EB34, ov_SC03_117 — counter/clamp variant; wave dj, corroborated by dk/dl/dm cards on the same function) L28015 - **ADDENDUM** — §37 — merged into the §153/§236-5 entry above (func_801A419C, wave dl) L28035 - **ADDENDUM** — §6 — merged into the §6 entry above (func_801811F0, wave dl) L28043 - **ADDENDUM** — to §238 (func_80182CB4, ov_SC02_000 — wave dl) L28051 - **ADDENDUM** — to §137a (func_801684B4, ov_MAIN_012; corroborated independently by func_80189E68, func_8017FF9C, func_801822B4, func_8018DA8C — wave dl) L28069 - **ADDENDUM** — to §8c / §88d (func_8016AB6C, ov_MAIN_012 — wave dm) L28085 - **ADDENDUM** — to §73 / §30#2 (func_800D1984, resident — wave dm) L28099 - **From** — ck + cl + cm L28227 - **ADDENDUM** — to §5a (func_80181F74, ov_SC03_112, wave cn) L28369 - **ADDENDUM** — to §265 (func_8017E26C, ov_SC04_016, wave cn) L28419 - **ADDENDUM** — to §42b (func_8018247C, ov_SC07_002, waves cu + cw) L28457 - **ADDENDUM** — to §195-E (func_800CFC1C, md_MAIN_003, wave cv) L28543 - **ADDENDUM** — to §215 addendum (func_800CB2C8, md_MAIN_033, waves cv + cw) L28588 - **ADDENDUM** — to §236 item 4 (func_8017D268, ov_SC04_006, wave cw) L28644 - **Harness-defect** — flags (not idioms — flagged for the operator) L28705 - **ADDENDUM** — to §250 (func_8017D7CC, ov_SC03_115 — cx/cy/cz/dr) L28746 - **ADDENDUM** — to §225 (func_8017E190, ov_SC03_115 — cx/cy/cz) L28791 - **ADDENDUM** — to §45-A (func_8017F6A4, ov_SC02_016 — cy/cz) L28844 - **ADDENDUM** — to §249 (func_80182ED4, ov_SC04_004 — dp/dr/dt) L28888 - **ADDENDUM** — to §199-A (func_801816FC, ov_SC02_005 — dp/dr/dt) L28933 - **Harness-defect** — flags L29022 - **§289** — an array local's address-taken base keeps every element's store alive, even though only one pointer escapes (`func_80189EFC`, ov_SC04_011) (P31 S60; waves #, byte-proven) L29104 - **ADDENDUM** — to §264-3 — the explicit entry copy's BOUND: required exactly when the RAW value must outlive the call that consumes only the PROMOTED value L29258 - **ADDENDUM** — to §276 — the SHARE direction: spell the second adjacent symbol RELATIVE to force ONE anchor L29290 - **ADDENDUM** — to §31's density-dummy dial (L2460/L2497) — the dose is TWO refs, and the dummy must sit where the loser is live-through L29306 - **§300** — S61 DISTILL BATCH NOTES (waves ab8/ab16/ab24/ab32 · g0a–g0f · m0a · ds1/ds2 · m0b) L29612 - **§306a** — T4 DISTILL ADDENDA (P31 S62; four byte-proven refinements to existing laws, verified against the book by an independent verifier; each names its parent §) L29911 - **Addendum** — §148-A2 addendum — the empty-asm insn_count dial (N bare `__asm__("")` = exactly +N real i (func_801835B0) L30463 - **Addendum** — Chained *2*2 array index folds to one copy;sll, not two (func_80011380) L30519 - **Addendum** — Last if-arm's nested if un-fuses the whole preceding chain (func_8017FE88) L30539 - **Addendum** — Orphan sh triplet to $sp is a write-only local array (func_8017F274) L30632 - **RETRIEVAL** — FAILURES this round (not new knowledge — a RETRIEVAL defect) L30666 - **Addendum** — Integer-space address arithmetic is a THIRD no-movable spelling, and a distant `SYM[0]` re (func_8017D89C) L30714 - **RETRIEVAL** — FAILURES this round L30775 - **§323b** — A SCRIPT THAT PARSES argv AT IMPORT CANNOT BE SHARED; EXTRACT THE PREDICATE, DO NOT COPY IT (P31 S67) L30945 - **§330** — THE NEIGHBOUR-SHAPE LEVER IS THE CHEAPEST FIRST MOVE, AND IT IS UNDER-USED (P31 S67; four independent instances in one 20-function wave) L31052 - **§347-addendum** — A THIRD INSTANCE, AND THE SHARPEST STATEMENT OF THE RULE (md_MAIN_025/func_800CB300, 243 ins) L31382 - **§359** — (§43-adjacent) — SPELL A SIGN-WIDEN AS AN EXPLICIT TWO-STEP, FUNCTION-SCOPED TEMP (P31 S68; byte-proven main/func_80012B58, 69 ins, 58 → 3 → 0) L31618 - **§360** — THE "COMPILER FOUND A SHORTER EQUIVALENT THAN THE TARGET" PAIR (P31 S68; main/func_800241C0, 68 ins, 68 → 19) L31631 - **§363** — ★★ — THE OVERLAY-LAYOUT ASSUMPTION IS A SYSTEMIC BUG CLASS, AND `main` IS THE EXCEPTION THAT FINDS IT (P31 S68; six instances, four of them in one session) L31715 - **§369** — REUSE THE **COMPARE CONSTANT'S OWN VARIABLE** FOR A MASK THAT KEEPS COALESCING (P31 S68; md_SC07_004/func_801AEC38, 365 ins) L31832 - **§372** — ★★★ — THE **COPY-CAPTURE PAIR**, AND THE ONE ZERO-BYTE EDIT THAT DEFEATS BOTH (P31 S68; byte-proven main/func_8003491C, 78 ins, fable escalation closed 5 → 0) L31949 - **CROSS-REFERENCE** — TO §370 — checked and found INAPPLICABLE here, which is the point L32028 - **§377** — THREE HARNESS DEFECTS FOUND IN ONE GATING SESSION, ALL "A CONFIDENT NUMBER ABOUT A SMALLER WORLD" (P31 S69) L32107 - **§380** — ★★★ — **A SECOND SET OF A PSEUDO DISQUALIFIES IT FROM `move_movables`** (P31 S69; main/func_800215F4, 465 ins, closeness 106 → 59 → 39) L32205 - **§382** — TWO FOLD REASSOCIATIONS THAT NEED THEIR OWN STATEMENT (P31 S69) L32242 - **§389** — ★★★ — `h_norm` IS BLIND TO INDEXED-GLOBAL RELOCS, SO FREE WORK BECOMES AN INVISIBLE SINGLETON (P31 S69; 31 stubs / 4,811 ins recovered, 8 banked same day) L32419 - **§390** — ★★★ — MINIMUM DISTANCE IS NOT MINIMUM WORK; RANK TWIN CANDIDATES BY EFFORT, AND FILTER LOOKALIKES BY RATIO (P31 S69; byte-proven ov_SC01_077/func_80184D50, banked) L32460 - **§394** — ★★ — TWO ALIGN-1 ACCESSES IN ONE FUNCTION RESERVE A PHANTOM STACK SLOT (P31 S69; ov_SC02_005/func_80180610) L32570 - **§314b** — ★★ — TWO ABS-RANGE GUARDS IN ONE FUNCTION NEED **DIFFERENT SPELLINGS** (P31 S69; byte-proven ov_SC04_002/func_8018691C, 111 ins) L32655 - **§397** — ★★★ — RE-RUN THE TWIN SCAN AFTER EVERY EXEMPLAR BANK; A CLUSTER SIBLING IS FREE THE MOMENT ITS EXEMPLAR LANDS (P31 S69; ~250k tokens spent proving it the expensive way) L32795 - **§398b** — ★★ — NAMING A SUB-EXPRESSION IN A LOCAL CHANGES WHICH PSEUDO SURVIVES; INLINE IT AT BOTH USE SITES (P31 S69; byte-proven ov_SC03_028/func_80183264, 93 ins) L32875 - **§399** — ★★★ — FOUR LEVERS FROM THE FINAL S69 ROUND (P31 S69; each byte-proven, none previously in the cookbook) L32902 - **§400** — ★★ — A BASELINE CHECK THAT CONFLATES "ABSENT EVERYWHERE" WITH "CHANGED UNDER US" SILENTLY DROPS NEW FILES (P31 S69; 8 files, one session) L32942 - **§401** — §401 L32977 - **§402** — §402 L33010 - **§403** — §403 L33034 - **§404** — §404 L33066 - **§406** — ★★ — THE PROLOGUE-WEAVE CLASS IS MECHANICAL AND LARGE (134 open stubs) L33168 - **§407** — ★★ — LATE-WAVE ADDENDA TO §405 (the last agents in) L33190 - **§409** — ★★★ — THE S71 JOURNAL-FUELLED WAVE: 100% FIRST-PASS MATCH, AND THE NINE LAWS IT BROUGHT BACK (P31 S71) L33271 - **§3-The** — nine laws this wave produced L33298 - **§413** — ★★★ — DIFFICULTY IS THE RESIDUAL CLASS, NOT `nins` — ROUTE THE MODEL TIER OFF HISTORY (P31 S71, Drew) L33483 - **§416** — ★★ — FOUR LEVERS FROM THE S71 OVERNIGHT LANE, none of which the cookbook held (P31 S71) L33582 - **§429** — ★★★ — EVERY HELD POINTER NEEDS ITS OWN LOCAL, AND A NEGATIVE-DISPLACEMENT BYTE STORE NEEDS ONE OF ITS OWN (P31 S72; `main/CdReadStateMachine`, MATCH 385/385) L34050 - **§434** — ★★★ — TWO SYMBOLS, ONE FRAME: RUN THE FRAME CHECK BEFORE DRAFTING ANYTHING LARGE (P31 S73; `main/SaveLoadRoutine` + `func_8002B0B4`, byte-verified) L34269 - **§438** — ★★★ — THE SAME-ADDRESS LEAD IS NOW SIZE-FILTERED: A HOMONYM IS WORSE THAN NO TWIN (P31 S74; ~12 of ~60 cards carried one) L34538 - **§441** — ★★ — THREE MORE INSTRUMENT DEFECTS FROM THE SAME SESSION, ALL OF WHICH BLAME THE SUBJECT (P31 S74) L34627 - **§443** — ★★ — A DERIVED DEPENDENCY MUST BE PROVISIONED BY *EVERY* PROVISIONER, AND WE HAVE TWO (P31 S69 → S74, the same defect twice) L34712 - **§445** — ★★ — `make clean BINARY=` IS FLEET-WIDE: THE VARIABLE IS ACCEPTED AND IGNORED (P31 S75) L34788 - **§448** — ★★★ — ASSEMBLY POSING AS C: 154 GAME FUNCTIONS THE REPORTS COUNTED AS DONE (P31 S75) L34908 - **§450** — ★★★ — REGENERATING A TARGET `.s` FOR A FUNCTION THAT IS NO LONGER A STUB (P31 S75; 146 functions unblocked, two silent defects caught by ONE cross-check) L35031 - **§451** — ★★★ — YOUR EVIDENCE HAS MORE THAN ONE SOURCE, AND THE ONE YOU QUERY IS PROBABLY THE WORSE ONE (P31 S75; 37 functions reclassified, 122 ins banked from one word) L35072 - **§452** — ★★★ — NOT EVERY VERBATIM BODY IS UNDECOMPILED WORK, AND §448'S HEADLINE OVERSTATED IT (P31 S75; 10-function burst, 0 banks, and the negative result is the finding) L35110 - **§462** — FOUR LEVERS FROM `main:func_80024054` (91 ins, 74/53/32 → 4) L35650 - **§480** — 🔴 — A STATIC BLOCKER CLASS THAT THE REAL PIPELINE ALREADY REMOVES IS A PHANTOM L35896 - **§454** — Rung D and rung R: taking a lever off a body that still has to compile to the same bytes (Phase 36 T6) L37610 ## All sections, in order - **§3-How** — to use this L30 - **§1** — Idiom catalog (asm pattern → C that produces it) L39 - **§3-I1** — Unsigned range check: `(x - lo) < (hi-lo)` → `addiu`+`sltiu` L41 - **§3-I2** — Byte mask forces `andi` even after `lbu` L47 - **§3-I3** — Division by a constant → magic multiply L54 - **§3-I4** — Runtime (variable) division → `divu` + zero-check `break` (NEEDS `--expand-div`) L60 - **§2** — Writing matching C (what makes gcc emit X) L69 - **§3-T1** — Loop pointer: top-of-body for `addu` induction, not constant-folded `addiu` L71 - **§3-T2** — Source statement order drives instruction scheduling L78 - **§3-T3** — Types L85 - **§3-T4** — Branch polarity: invert the source condition to flip gcc's chosen branch L90 - **§3** — When a diff is pure scheduling → decomp-permuter (harness built, Phase 6) L107 - **§3a** — Escalation TIER above the permuter — web-research the compiler internals (HIGH VALUE, proven) L128 - **§3b** — §31-directed permuter mutation — bias the search over the class's levers (Phase 24 T5) L147 - **§4** — Flag/toolchain gotchas L190 - **§5** — Known hard-residual classes (instruction-identical, one byte-exact blocker) L199 - **§5a** — Cross-jump tail-merge — gcc collapses two byte-identical blocks the original kept separate (FIX FOUND) L211 - **§6** — Per-module optimization mixing — the -O0 boot module (Phase 7) L265 - **Detecting** — the opt level (do this first) L273 - **Build** — mechanism — per-file opt override (splat resegmentation) L307 - **§7** — PsyQ SDK types & symbols (the library-call prerequisite) L322 - **§8** — rodata island (compiler jump tables) — the `.data→.rodata→.data` sandwich (Phase 7) L339 - **§8a** — rodata island in a flat OVERLAY — the tail sandwich, per matched jr-function (Phase 26 — PoC PROVEN) L361 - **§8a-pad** — a trailing `.word 0x00000000` under a jtbl dlabel is `.align` PAD, not an entry (Phase 26 session 6, byte-proven) L399 - **§8b** — MULTI-jtbl per overlay — the `ld_interleave --order` sandwich + the same-subseg cases (Phase 26 session 4) L423 - **§8c** — Splitting a TU means rebuilding its DECLARATION ENVIRONMENT, not moving text (Phase 26 session 6) L456 - **§8d** — Templating a body INTO a TU must not CHANGE its declaration environment — demote the carried data externs (Phase 26 session 8, byte-proven on `func_8015AE2C` ×133) L502 - **§8e** — The jtbl ALIGNMENT LAW + the pad-spec filter — multi-table .rodata spans (Phase 29, byte-proven; `.run/probe_jtbl/verdict.md`) L549 - **§9** — Link real PsyQ library objects byte-exact (Phase 7 — GO proven) L632 - **§9.1** — Generalised per-object linker — `tools/psyq_link.py` (+ `psyq_link_lib.py`), 18/18 libcd byte-exact L661 - **§9.2** — Wire a library region into the build with NOLOAD — no data carving (`tools/psyq_link_region.py`) L685 - **§9.3** — Make it the build: resegment + swap + resolve (`tools/psyq_integrate.py`, libcd DONE) L704 - **§9.4** — Integrating a SECOND library (libgs block 6 after libcd) — multi-library gotchas L731 - **§9.5** — Integrating a WHOLE multi-block library in one call (full libgs — Phase 7 session G) L759 - **§9.6** — Scaling library linking to the whole EXE (Phase 8 — 8 libs linked, 20%→50% byte-identical) L785 - **§9.7** — Binary-agnostic toolchain refactor (Phase 9) — the reusable pattern for Gen2 L821 - **§10** — Closing the regalloc/scheduling hard tail by hand (LZSS, Phase 7 session F — the full close) L854 - **§3-The** — clean object-level metric (use this, not the permuter score, for jtbl/rodata functions) L863 - **Residual** — A — commutative `|`/`&`/`+` result lands in the wrong source-operand register L875 - **Residual** — B — a `return ` materialised late / merged instead of distributed per-site L886 - **§11** — Cross-binary dedup & code-sharing (Phase 11 — "one match unlocks many") L927 - **§3-The** — mechanism: game-code dedup is SOURCE-LEVEL, not an object swap (R-D1, the key lesson) L945 - **Per-binary** — toolchain provenance (R24) L978 - **§12** — Ultracode harvest — parallel-draft + byte-gate at scale (Phase 12, resident: 1.4%→71.7% in one session) L983 - **§13** — Add a location overlay — the canonical runbook (Phase 13; the Phase-15 fleet recipe) L1075 - **One-command** — onboarding — `tools/new_overlay.sh ` L1081 - **§3-The** — flat-blob overlay config (what the template encodes) L1090 - **§3-THE** — NON-4-ALIGNED-OVERLAY GOTCHA (≈75% of the fleet; fixed in the template + Makefile, automatic) L1098 - **§3-The** — A→B→C per-overlay workflow L1110 - **Dedup-credit** — (§11) for overlays — two shapes L1127 - **Fleet** — build — `make build-all` / `make check-all` L1139 - **§14** — Propagate a matched function across the fleet — `tools/dedup_propagate.py` (Phase 15) L1145 - **§14a** — The fleet bulk run (Phase 15 — `--auto-from`, 947 REAL → 74,527; 3.82% → 22.14% in one pass) L1192 - **§14b** — Harvesting the UNMATCHED shared core — the match_one wall (Phase 15) L1215 - **§14c** — Callee-signature-aware harvest — breaking the extern-type-conflict wall (Phase 15, T6) L1236 - **§14d** — Deterministic recovery beats agent waves on the hard tail (Phase 15, the final-session finding) L1292 - **§14e** — Two hard-tail dead-ends (Phase 15 — documented so they aren't re-attempted) L1321 - **§15** — Struct-heavy shared-core pipeline (Phase 16) — empirical determinations (S0) L1336 - **§16** — Guided hand-matching the struct-heavy core (Phase 17 — beats the §15 brute-force) L1358 - **§17** — The compiler-quirk wall — the matching TOOLKIT (Phase 18; gcc-2.7.2 source + byte-gated) L1399 - **Register-allocation** — ORDER (call-crossing $s0/$s1 swap) → FORCE it with register pins (byte-proven) L1407 - **§3-The** — STEERABLE idioms (byte-confirmed this phase) L1427 - **§3-The** — pipeline gotcha — match_one ≠ the gate; `sig_unify` is MANDATORY L1444 - **§3-The** — LOOSE-TYPING wall is real for narrow params (Phase 16, reconfirmed) L1451 - **Strategic** — conclusion — the match-% lever post-research L1458 - **§17a** — The TOOLKIT at WAVE scale (Phase-18 Step-3b/Step-1 — measured) + the pipeline-integration gotchas L1468 - **§18** — Per-file `-O0` split inside an overlay/blob (Phase 19 T1) L1542 - **§19** — Scaling the toolkit waves — the recovery PIPELINE + the propagation CAP (Phase 19 T3) L1620 - **§20** — The wave-at-scale GATE CAP + residual-class verdicts (Phase 20) L1655 - **§3-THE** — CAP: at this tail the match_one→gate gap is the LOOSE-TYPING CALL-GRAPH wall (not cheap plumbing) L1660 - **Diagnostic** — lesson: a failed in-TU build leaves a STALE `.o` L1677 - **§3-NEW** — / CONFIRMED residual classes (this session) L1683 - **Operational** — gotchas (cost real time) L1704 - **What** — WORKED — the Phase-20 reusable wins L1712 - **Phase-20** — RESOLUTION — `tools/cast_call_sites.py` BUILT + the cap re-diagnosed (R14, byte-proven) L1724 - **§21** — wave-distilled idioms (Phase 21) L1775 - **§22** — DEF-side loose-typing recovery + grinder blacklist (Phase 21) L2028 - **§23** — Giant `func_80153E00` cracked (scalar-data CAST); BUT most giant drafts are STALE+INCOMPLETE, and a diag MUST remove artifacts (Phase 21 — byte-proven + a self-correction) L2055 - **§24** — The `ov_SC01_077_a.c` split-file vein: split-aware propagation, but loose-typing-gated (Phase 21) L2108 - **§25** — The "schedule" class is mostly COALESCING (pin-crackable), not scheduling; + the gate two-stage + h_exact over-counts ×134 (Phase 21, cont.6) L2155 - **§3-The** — crack: the residual was gcc-2.7.2 COPY-COALESCING, not the instruction scheduler L2167 - **§3-The** — genuine scheduler — `rank_for_schedule` (sched.c), for when it IS scheduling L2183 - **§3-The** — genuine schedule WALLS (do NOT re-grind — stub) L2192 - **§3-h** — _exact OVER-COUNTS ×134 — verify shareability before crediting a class's "reach-134" (R14) L2200 - **§3-The** — gate two-stage — sig_unify is a FALLBACK, not unconditional (`gate_stage.py`, §19 folded in) L2212 - **§26** — The cheap close=0 recovery lever is EXHAUSTED; `idiom_loop --assess` was DOUBLY inflated (Phase 21, cont.7) L2222 - **§3-The** — `_a` close=0 recovery banks 0/20 — same def-side wall as MAIN (0/40) L2228 - **Where** — this leaves the reach-134 tail (cont.6 option-3, now CONFIRMED byte-backed) L2256 - **§27** — Giant matching recipe (Phase 21 cont.7b — validated on func_80176D94, 152 ins) L2282 - **§28** — Banking a "close=0 gate-rejected" giant: the canonical-extern recovery (Phase 22 T1, byte-proven on func_8015126C ×134) L2348 - **§28a** — decomp.wiki GCC patterns worth trying on BFM giants (decomp.wiki/compilers/GCC, raw at decompals/decompedia; PS1-applicable subset) L2368 - **§28c** — The close=0 recovery is NOT fully exhausted (§26 corrected, R14); + the dedup_propagate registry-skip recovery (Phase 22 T2) L2375 - **§28b** — The struct-walled close=0 giant (§28 case #3) IS bankable: the engine_types.h type-lift (Phase 22 T2, byte-proven on func_80156B74 ×134) L2380 - **§28d** — The "macro-extern-injection" lever: freeing reach-134 inline matches dedup_propagate skips as "not self-contained" (Phase 23, `tools/inject_capped_externs.py`) L2389 - **§29** — Reasoning-model (GLM5.2) DEF-side reconciliation idioms + the wall's hard limit (Phase 23 T10.7, `tools/glm_reconcile.py`) L2397 - **§30** — Fable5Max cracks §20/§10 "unsteerable" from the gcc SOURCE: store-vs-load is a `/s` aliasing flag, the def-side wall has a macro escape, + the birthing-boost (Phase 23, byte-proven on giant `func_8014EE14` 248 ins ×134) L2405 - **§30a** — §30 generalizes to the FULL near-miss backlog (via STANDARD Opus agents, not just Fable5) + 2 more steer levers + the mechanical-integration throughput unlock (Phase 23 (a)) L2417 - **§31** — THE gcc-2.7.2 CODEGEN MAP: pass → residual → C-lever catalog (Phase 23; 4 Fable5 agents read the compiler source) L2425 - **§32** — The region-a CAMERA-GIANT idiom set: struct-base hoisting + 4 sibling levers (Phase 24 T7, Fable5-cracked on `func_80129CF8` 191 ins, match_one MATCH; transferable to the 6 sibling giants) L2445 - **§33** — Automating the giant decl-reconcile: `tools/reconcile_decls.py` (the DATA analog of §20's `cast_call_sites`) + a fleet-majority type oracle (Phase 24 T7b, byte-proven on `func_80129CF8`) L2457 - **§34** — The `func_80138ED0` giant crack: gcc-2.7.2's **3-qty sort bug** + the **zero-byte asm allocation toolkit** + the **giv-init fence** (Phase 24 T5; Opus→close=21, Fable5→MATCH ×134) L2477 - **§35** — The region-a sibling-giant harvest: difficulty ≠ $s-reg count (it's global-array hoisting) + the banking recipe + new loop idioms (Phase 24 T7, 5 parallel Opus-Max agents) L2494 - **§36** — Fable5 giant-crack levers (Phase 24 T7 Fable5 batch; accumulates as each lands — full RTL dumps in `.run/t7/fable/`) L2510 - **§37** — The T7 §G giant endgame: 6/8 cracked, meta-laws + transferable levers (Phase 24, 2026-07-07; FULL byte-verified detail + gcc-2.7.2 line cites in `docs/gcc-2.7.2-map/t7g-giant-harvest.md`) L2529 - **§38** — The WHALE `func_80144B9C` (770 ins): the -O0 struct-assign memcpy idiom + the -O0 reach-134 ×134 rollout (Phase 24 T7 §G, cheap Opus — no Fable5, no calls.c) L2550 - **§39** — The ×1→×134 giant-endgame: propagate a matched **-O2** giant via the NATIVE DEFINE-macro path (Phase 24 T7 §G close, 2026-07-08) L2560 - **§40** — Structural families: the MECHANICAL symbol-remap (crack one exemplar → remap the rest, ~0 tokens) (Phase 25 T3/T7, 2026-07-08) L2576 - **§40a** — The DECL-RECONCILE pass (type-lift so remapped drafts compile in the sibling TU) (Phase 25 T7.2, 2026-07-08) L2624 - **§40b** — The reloc-tracker blind spot: the indexed-global idiom that hid the "reach-1 tail" (Phase 26 Task 1, 2026-07-11, byte-verified V0/V1) L2657 - **§40c** — The h_seq per-sibling reconcile: templating a reconcile-class crack ×134 (Phase 26 Task 8, 2026-07-12, byte-proven) L2699 - **§31-triage** — R17 applies to CODEGEN residuals, never to a compile ERROR (Phase 26 session 7, Drew asked) L2716 - **§41** — The DEF-SIDE canonical-sig wall: mechanically banking a drafted giant past `conflicting types` (Phase 25 T5b batch-2, 2026-07-09; `tools/canon_sig_reconcile.py`, byte-proven on `func_8013B274`) L2733 - **§41a** — v3.1: the def-side wall was ~71% TOOL-shaped — the five measured defects + the laws that dissolve them (Phase 25 T6, Fable5, 2026-07-09) L2792 - **§41b** — T7 execution: the object-only probe OVER-counts BANKABLE by two link/rodata classes (Phase 25 T7-M1, 2026-07-10) L2847 - **§41c** — T7-M2: the ×134 def-side-wall sweep via per-sibling RE-reconcile (Phase 25, 2026-07-10; 4,389 banks, ~0 agent tokens) L2877 - **§41b** — addendum — M4 "reconcile_decls" tier is ALSO a probe over-count: 0/8 mechanical (Phase 25 T7-M4, 2026-07-10) L2896 - **§41d** — `void`→`s32` is NOT always byte-neutral: gate the RAW draft FIRST (Phase 26 session 6, byte-proven) L2916 - **§42** — The F-band ≤28 regalloc crack wave: register-pin/DENSITY levers beat the permuter (Phase 25 T7 F-band, 2026-07-10; Ultracode 9-worker wave, 4/9 banked byte-identical, 266 swept ×134) L2938 - **§42a** — addendum — wave 2 (residuals + 29-100 band): iso-MATCH ≠ real-TU bank, the memcpy→struct-assign fix, +5 levers (2026-07-10b) L2991 - **§42b** — addendum — wave 3 (Max, 2026-07-10c): THE STALE-OBJECT GATE TRAP + the read-global `&`-cast drift + fix L3032 - **§42c** — addendum — wave 3 (Max orchestrator + CORRECTED Ultracode fan-out, 2026-07-10c): the real-TU-faithful parallel harness (rtu_match) + 7/9 crack, ZERO iso-drift L3071 - **§42d** — addendum — wave 4 (rtu_match fan-out over the mapped frontier, 2026-07-10c): 24/26 MATCH, +5 durable levers L3120 - **§42e** — propagating a CRACK ×134: the def-finder bug + the byte-drift residual (the "remap-fail" misdiagnosis) L3159 - **§43** — The K&R s16-param definition DISSOLVES the "narrow-param wall" for by-value register args (Phase 25 task A, Fable5 crack of the 369-ins giant `func_80166994` ×134, 2026-07-11) L3233 - **§44** — The Phase-25 cheap-Opus giant batch: 5 structural levers + the §43 extension (2026-07-11, 6 crackers over the frontier giants) L3280 - **§45** — The flagship `func_80133CD4` crack (399 ins ×134): the merged-variable permutation-breaker + the 1-death local-alloc gate (Phase 25 task A giant escalation, Fable5 gdb-on-cc1, 2026-07-11) L3336 - **§46** — The `func_80178D40` crack (890 ins ×134, the heaviest core in the game): four LOOP-STRUCTURE levers cheap-Opus found by reading loop.c/jump.c/cse.c (Phase 26 session 8, 2026-07-13) L3351 - **§47** — The live-length SLIDER: splitting a global.c allocno-priority TIE with one zero-byte asm (Phase 26 session 8, Fable5 Max, byte-proven on `func_8017BEBC` 952 ins ×113) L3403 - **§48** — The 12-core jr crack wave: the ALLOCNO-PRICING dials and the EBB rule (Phase 26 session 8, Ultracode, 9/12 MATCH first pass) L3443 - **§3-A.** — ALLOCNO-PRICING DIALS — move a value into the register you want, byte-neutrally L3451 - **§3-A4** — SINK THE CONSUMER CALL INTO THE ARMS (the inverse of A1; `func_8016AB6C`, byte-proven) L3482 - **§3-B.** — THE EBB RULE — the general form of §46-L2 L3506 - **§3-C.** — TYPE- AND SHAPE-DRIVEN CODEGEN (the C type literally selects the addressing mode) L3522 - **§3-D.** — THE CROSS-JUMP RATCHET (the sharpest new trap — `func_80131340` L-C) L3546 - **§3-E.** — Meta L3556 - **§49** — The LUID DIAL: a zero-byte SCHEDULING dial (the sched.c analogue of §47) — `func_8017A4AC` (536 ins ×134), Phase 26 session 8 L3565 - **§50** — Refinements that BOUND §47/§48 (from the `func_80135EB0` wall, 21→6; Phase 26 session 8) L3614 - **§51** — TOOLING INTEGRITY: the silent skip, and how to hunt it L3668 - **§51a** — The bug class L3674 - **§51b** — Why the byte-gate cannot save you L3690 - **§51c** — THE METHOD (do not audit by reading the regex) L3700 - **§51d** — THE LAWS L3715 - **§51e** — The false-wall pipeline (why this is not just hygiene) L3777 - **§51f** — Checklist for any new corpus-scanning tool L3793 - **§51g** — When the thing you are scanning has a GRAMMAR, parse the grammar (`tools/cdecl.py`) L3807 - **§52** — The WALKER-FAMILY skeleton: 6 regalloc-order levers + a deeply-characterized intrinsic wall (`func_80178004`, 165 ins ×134; Phase 26, Fable5, 2026-07-15) L3951 - **§3-The** — 6 levers (the "walker-family skeleton" — apply to the 12 siblings; levers 1-5 retire ~half the residual) L3962 - **§3-Why** — the wall is (probably) intrinsic L3983 - **§52a** — The regalloc sibling wave: new levers + two new wall classes (cheap-Opus applying §52, 2026-07-15) L3997 - **§52b** — Sibling wave 2: more de-pin levers, a third wall class, and the match_one→whole-binary gap at scale (2026-07-15) L4035 - **§53** — SWEEP A FAMILY WITH THE TOOL ITS EXEMPLAR NEEDED: the jr/switch carve, and how omitting it manufactured the "families don't template" doctrine (Phase 28 T1, 2026-07-15) L4066 - **§3-The** — case L4071 - **§3-Why** — 0/8 was structural, and predictable from two words L4083 - **§3-The** — rule L4100 - **§3-The** — meta-lesson (R35, and why this one is expensive) L4116 - **§55** — Core-crack wave levers + the GATE-ORCHESTRATION law (Phase 29 T3, 13-agent ultracode wave, 2026-07-17) L4131 - **§55a** — New byte-proven levers (each from a banked or near draft) L4136 - **§55b** — THE GATE-ORCHESTRATION LAW (3 traps, ~3.5h lost; all recovered, 0 data lost) L4160 - **§55c** — Sizing the propagate: a TARGETED propagate is ~4 min/core, and "it's slow" was a BROKEN-TREE ARTIFACT L4179 - **§54** — `--fix-def-sig`: the member's CANONICAL DECLARATION is a build step too (tiny-IMM mega-pools, +4,801, Phase 29 T6, 2026-07-16) L4201 - **§56** — Banking a hand-drafted GIANT into its exemplar TU: self-contained draft vs live-TU decls, 4 reconciliation tactics (`func_8013FAF8`, 312 ins ×138, Phase 29 T4, 2026-07-17) L4225 - **§56b** — PROPAGATING an h_seq giant: the exemplar's externs MUST be fleet-canonical, not the draft's types L4270 - **§57** — The SELF-decl normalize: the sibling's OWN caller declares the templated fn divergently (the third §17a-1 direction; `tools/normalize_self_decls.py`, func_801670E4 ×137, Phase 29, 2026-07-18) L4294 - **§57a** — Two NSD corrections + the SURGICAL-ONLY law + the honest broad-sweep yield (Phase 29, 2026-07-18) L4341 - **§58** — match_one MATCH ≠ BANK: the four blind spots + the crack-wave reconcile-before-bank law (Phase 29 crack-wave, 2026-07-18) L4381 - **§59** — Three h_seq sweep-residual classes match_one/the-exemplar-bank don't reveal (Phase 29 crack-wave close, 2026-07-18) L4415 - **§60** — Classify the residual, don't rank it: the deterministic residual→class classifier and what it measured about the backlog (Phase 29 Task-13A, 2026-07-21) L4445 - **What** — it measured — the whole open backlog, byte-grounded L4469 - **§3-Two** — corollaries worth remembering L4488 - **§3-The** — parallel-probe race this surfaced L4501 - **§60a** — What the first DIRECTED grinder run exposed (Phase 29 Task-13B, 2026-07-21) L4510 - **§60b** — The plateau autopsy's verdict: a `partial` drift is a WRONG DRAFT, not a missing transform (Phase 29 Task-13B close, 2026-07-21) L4548 - **§61** — Task 14: the gate ladder's missing stage is the ARITY pre-pass, and it is TU-side not draft-side (Phase 29, 2026-07-21) L4583 - **§61a** — The Task-5 wave: 11/12 MATCH, 0 banked — three DISTINCT integration walls, each now named (Phase 29, 2026-07-21) L4644 - **§61b** — The jtbl gate stage: built, and the ORDERING law it exposed (Phase 29 Task-14 stage 4, 2026-07-21) L4696 - **§61c** — The jtbl bank is INCREMENTALLY valid and CLEAN-INVALID (Phase 29, 2026-07-21) — the blocking finding L4751 - **§61d** — The undo was eating the tree: two tools, one defect, invisible to the byte-gate (Phase 29, 2026-07-22) L4817 - **§62** — The jtbl RECONCILE must also follow the carve: the post-carve draft reconcile (`harvest_verify._jtbl_reconcile`, Phase 29 SESSION-11, 2026-07-22) L4871 - **§63** — The fresh-138 DEF-SIDE blocker: fix the HEADER decl, not the draft (`fix_header_decl.py`, Phase 29 SESSION-13, 2026-07-23) L4915 - **§64** — The §20 type-lift's three laws: fold the tagged typedef, check VISIBILITY, and strip only what is TEXTUALLY IDENTICAL (`lift_types.py`, Phase 29 SESSION-14, 2026-07-23) L4946 - **§64a** — VARIANT types: UNIQUIFY the camps, do not reconcile them (`uniquify_type.py`, Phase 29 SESSION-14) L4995 - **§63** — UPDATE (Phase 29 SESSION-14) — `fix_header_decl`'s "SAFE" verdict is FLEET-BLIND; it MUST be R22-validated L5038 - **§65** — The stranded-draft recovery: BLAST-RADIUS TIERS, and the per-overlay de-macroize that refutes §20's DEF-conflict wall (`tools/demacroize.py` + `tools/blocker_probe.py`, Phase 29 SESSION-16, 2026-07-24) L5053 - **§65a** — The blast-radius taxonomy (makes §61's law structural instead of remembered) L5068 - **§65b** — The escape: de-macroize the instantiation, don't touch the shared header L5083 - **§65c** — `rtu_match` MATCH → bank held 13/13 on self-decl, but broke on the FIRST callee-decl case L5119 - **§65d** — Existing-ladder baseline, measured (do this before building a recovery stage) L5129 - **§65e** — Two oracles, and the disagreement is the finding (R34 in practice) L5139 - **§65f** — The de-macroize lever's BOUNDARY, measured: byte-neutral 14 times of 15, and the 15th is why the gate exists L5156 - **§65g** — Where the cheap levers STOP: the local-type and in-TU-self-decl classes did not yield L5177 - **§66** — Exercise a banking driver's SUCCESS path before pointing it at a wave: the free re-bank test (Phase 29 SESSION-17, 2026-07-24) L5197 - **§66a** — The widest write in a pipeline is the one most likely to be UNDECLARED L5219 - **§66b** — A metric parsed out of another tool's prose goes NULL silently when the label changes L5238 - **§66c** — Before a wave, verify the FUEL exists; an "already attempted" set built from the wrong directory lies in BOTH directions L5255 - **§66d** — The permuter⇄reader loop: alternate a random search with a byte-verified idiom, and let `residual_class` decide whose turn it is (Phase 29 SESSION-17) L5283 - **§66d-1** — What transfers between giants is the LOOP, not the PIN L5318 - **§66d-2** — Two operational sharp edges L5327 - **§66d-3** — Read the ILS per-cycle SERIES, not its final best: a repeated score and a still-falling one look identical in a summary line and mean opposite things L5338 - **§67** — The arg-copy PLACEMENT lever: launder a parameter into a fresh pseudo AT the statement where the target's copy lands (Phase 29 SESSION-18, `func_8014D820` 25 → 16) L5353 - **§67a** — Run the symbol-set guard BEFORE you pay for a gate (`tools/symcheck.py`, Phase 29 SESSION-18) L5443 - **§66d-4** — "ILS converged" means converged FOR THAT WEIGHT PROFILE, not a floor (amends §66d-3; Phase 29 SESSION-18) L5476 - **§66d-5** — `residual_class`'s "structural ⇒ permuter CPU is waste" is WRONG for schedule permutations (measured, Phase 29 SESSION-18) L5529 - **§68** — A comment-only line halted the extern scan, and the skip label blamed the type cap (Phase 29 SESSION-18) L5562 - **§69** — How to attack a behemoth: map it, don't draft it (Phase 29 SESSION-18, `func_80183814`, 5,122 ins) L5607 - **§70** — The giv-init base register: walk the PARAMETER, not a copy of it (Phase 29 SESSION-18, `func_801777BC`) L5657 - **§71** — Before mapping a giant, look for an already-matched SIBLING beside it (Phase 29 SESSION-18, `func_8017D960`, 3,338 ins) L5698 - **§72** — A `register __asm__` pin is a PREFERENCE, not a reservation (Phase 29 SESSION-18, `func_8017F510`) L5764 - **§3-The** — honest fix was source-level and cheap L5788 - **§3-Giv** — record order (the §70 family) L5798 - **What** — is left, and what is byte-recorded as SPENT L5804 - **§73** — A def-side self-decl conflict has TWO axes: RETURN (fleet widen, T2) and PARAMS (casts at each use, T0). Diagnose which before paying for the expensive one (Phase 29 SESSION-19, `func_8014F3E8` + `func_8014D4C0`) L5821 - **§74** — Auditing a pinned draft: the §72 hazard is CALLER-SAVED pins spanning a call, and only the disassembly can tell you (Phase 29 SESSION-19, `func_8017D960` b2, 5 pins) L5860 - **§75** — A propagation cap is usually a MINORITY-SPELLING SOURCE OVERLAY, not a wall: census the carried extern before believing the exclusion message (Phase 29 SESSION-19, `func_8014F3E8` ×4 → ×138) L5904 - **§75a** — The exclusion classes, enumerated with named causes (Phase 29 SESSION-19, the 134-binary `dedup_extend` sweep) L5963 - **§75b** — A body's preamble can carry `#define`s, not just `extern`s; extraction lifts only the externs (Phase 29 SESSION-19, `func_80165CA0` ×3 → fleet) L5986 - **§75c** — Class-B's remedy is the FULL §17a-1 PAIR (decl **and** call-site cast); a decl-only fix moves the error and looks like a new wall (Phase 29 SESSION-19, `func_8012F14C`) L6035 - **§76** — The allocno CLASS (local vs global) is the dominant regalloc lever, and C reaches it ONLY through declaration scope and variable reuse (Phase 29 SESSION-19, behemoth #3 `func_8017F510` 1,511 ins, 97 → MATCH, pin-free) L6066 - **§3-The** — mechanism, with citations L6075 - **§3-The** — attribution primitive (use this before calling anything a scheduling residual) L6095 - **§3-Two** — diagnosis traps this function proved L6102 - **Practice** — Practice L6112 - **§77** — Every extraction tool carries a NARROW hard-coded preamble set; anything outside it silently caps the body's reach. Diff the preamble before you gate. (Phase 29 SESSION-19 — three variants in one session, two different tools) L6122 - **§3-The** — CANDIDATE gate and the REAL gate need DIFFERENT preambles — keep the difference out of the bank L6162 - **§78** — A LENGTH drift can be a register grant in disguise; and `fold` never leaves a literal first in an `|` chain (Phase 29 SESSION-19, behemoth #2 `func_8017D960` 3,338 ins, 1806 → 0, pin-free) L6212 - **§3-The** — drift was an allocation decision, not missing code L6218 - **§3-The** — economics L6249 - **§79** — For a 0-callee giant, fingerprint by DATA symbols (§71 cannot fire); and the STACK-SLOT ORDER is a declaration-order oracle (Phase 29 SESSION-19, `func_8017BF14` 4,763 ins, cold start → 45/4763) L6258 - **§71** — has a blind spot, and this is it L6264 - **§3-NEW** — LEVER — the frame layout reads back the original declaration order L6278 - **§76** — confirmed at scale, and a pin nuance L6289 - **§3-The** — residual, and the honest read L6298 - **§80** — A do-not-re-buy entry is scoped to its BASE, not to the function; and the pin's hidden cost is an unconditional `qty_phys_sugg` (Phase 29 SESSION-19, `func_8017BF14` 45 → 0) L6308 - **§3-The** — pin's hidden cost, with the citation L6329 - **§3-The** — flagged "#1 move" LOST — and why the failure is informative L6340 - **§78** — 's attribution primitive, run and reproduced L6350 - **Cold-start** — economics, now complete L6356 - **§81** — Banking a jr (jump-table) function: the 3-step carve chain, and why `match_one` cannot see the problem (Phase 29 SESSION-19, `func_8017C954`) L6361 - **§3-The** — defect this exposed: a shared type that is present but invisible L6392 - **§82** — Two source-shape oracles from behemoth #6: a duplicated `addiu $aN,$sp,K` across a `jal` means the block was INLINED, and scalar-vs-aggregate decides WHEN a stack slot is allocated (Phase 29 SESSION-19, `func_8017C730` 1,061 ins) L6408 - **§3-1.** — The inlined-helper signature L6413 - **§3-2.** — Scalar vs aggregate decides *when* the slot is allocated L6427 - **Also** — reproduced on this function L6439 - **§3-And** — the banking footnote (§75a class A, one line) L6443 - **§83** — The parameterised-repeat law, the spill-area trap, and why a per-case edit cannot move a per-case symptom (Phase 29 SESSION-20, `func_80183814` 5,122 ins, cold-ish start → 36 structural / 99.3%) L6450 - **§83a** — READ THE HEADLINE NUMBER CORRECTLY: a LENGTH drift makes `match_one`'s count meaningless L6457 - **§83b** — THE LEVER: find the parameterised REPEAT before decoding case-by-case L6465 - **§83c** — TRAP: a "dead local" in a prior draft may be gcc's OWN spill area L6483 - **§83d** — CSE's quantity budget is WHOLE-FUNCTION, so a local rewrite cannot fix a local symptom L6490 - **§83e** — §80 vindicated again, on the same day it was written L6504 - **§84** — The DERIVED-OFFSET remap bug: a hand-computed literal that encodes the DISTANCE between two per-overlay symbols, and why `match_one` is structurally blind to it (Phase 29 SESSION-20, `func_8013D53C`) L6510 - **§3-The** — construct L6515 - **§3-Why** — it survived every candidate gate L6533 - **§3-THE** — FIX IS MECHANICAL — the tool already holds the answer L6548 - **Scope** — , measured (do not over-generalise — §80) L6556 - **§3-Two** — ladder lessons banked with it L6566 - **§85** — The RETURN-axis fleet widen is ALL-OR-NOTHING: widening the shared header alone guarantees a conflict in the source overlay (Phase 29 SESSION-20, `func_8012CC88` / `func_8014D12C`) L6577 - **§3-The** — conflict L6582 - **§3-THE** — FAILURE MODE — widening only `engine_core.h` is worse than not starting L6590 - **§3-The** — precondition, and how to check it in one grep L6597 - **§3-Do** — the WHOLE axis in one edit, then R22 once L6605 - **Reading** — , for the next person L6614 - **§86** — Pinned-exemplar templatability is a PER-FAMILY property, not a per-member rate; and the §42e pin guard is now over-conservative (Phase 29 SESSION-20) L6627 - **§3-The** — guard refuses a class that largely works L6629 - **§3-THE** — LAW: all-or-nothing PER FAMILY L6638 - **§3-Why** — the two live families differ from the three dead ones — the open question L6665 - **§87** — `match_one` COMPILES but never LINKS, so an unresolvable data symbol reads as MATCH; and stored drafts go STALE against the tree (Phase 29 SESSION-20) L6671 - **§3-The** — blindness ladder, now complete — FOUR classes `match_one` cannot see L6687 - **Consequence** — for the backlog ledger L6697 - **§3-The** — cheap discriminator, before spending a sweep L6704 - **§88** — `cross_jump` will not merge a common suffix containing a CALL; and §78 is scoped to ORDERED comparisons only (Phase 29 SESSION-20, the behemoth close-out) L6712 - **§88a** — repeated CALL-shaped blocks are left UNMERGED; call-free tails are merged for you L6717 - **§88b** — the `slti` literal-position law (extends §78 to comparisons) L6724 - **§88d** — BANKING ORDER: run the §81 carve chain BEFORE banking, never after L6739 - **§88e** — a wrong diagnosis, refuted properly (the model for how to treat an inherited lever) L6746 - **§88f** — the missing rung: a RELOCATION gate between `match_one` and the binary L6756 - **§89** — Two throughput rules the project already had written down and was not following (Phase 29 SESSION-20) L6764 - **§89a** — MEASURE the write set; do not assert its tier (`tools/blast_radius.py`) L6770 - **§89b** — the parallel gate farm existed; the family path could not reach it (`tools/sweep_parallel.py`) L6783 - **§3-The** — standing sequence L6796 - **§90** — Five tool-integrity laws from one session, each of which changed an answer (Phase 29 SESSION-21, 2026-07-27) L6804 - **§90a** — A comparison tool MUST share its reference oracle's index space, exactly L6810 - **§90b** — "Byte-neutral" is not "wanted": undo on the SUCCESS path too L6827 - **§90c** — A library-callable function must FAIL CLOSED on an unconfigured module L6837 - **§90d** — Do not measure a live wave's drafts (§87 in real time) L6848 - **§90e** — An agent's CONCLUSION and its EVIDENCE fail independently — re-derive the premise, not the fix L6856 - **§91** — A structure-TRANSFER is only valid where the structure corresponds: the `--like` role trap (Phase 29 SESSION-21, `func_8012AAAC` ×137) L6880 - **§3-The** — three-hypothesis trail, because two of them were wrong and the wrongness is instructive L6909 - **§92** — Conforming a DECLARATION: pointer changes are caller-neutral, scalar-WIDTH changes are not (Phase 29 SESSION-21, `tools/conform_decls.py`) L6929 - **§93** — `set -o pipefail` attributes a pipeline failure to the LAST stage, not the failing one (Phase 29 SESSION-21, `func_8014D820`) L6960 - **§94** — A family sweep's `0/N` is a TYPE-CARRY failure until proven otherwise: lift the exemplar's local types, ALL of them, transitively (Phase 29 SESSION-21, `func_8016B6BC` 0/137 → 137/137) L6985 - **§95** — `reconcile_tu` dropped the SIBLING declarators of a multi-symbol `extern` line (Phase 29 SESSION-21, `func_80176218`) L7021 - **§96** — The same rewrite, one shape down: `reconcile_tu` matched statements to lines by TEXT, so every COMMENTED declaration was silently skipped (Phase 29 SESSION-22, `func_80176218` banked) L7056 - **§97** — The gate's own tree hygiene: a refused carve, an unchecked recovery, and a snapshot that captured a dirty tree (Phase 29 SESSION-22) L7104 - **§98** — `conform_decls` had three defects, and only the third needed R22 to find (Phase 29 SESSION-22, `func_8014CF04`) L7151 - **§99** — The narrow-param wall is a DEF-side problem with a ZERO-blast-radius fix: convert the definition to K&R (Phase 29 SESSION-22, `func_80175AB8` + `func_80175DA8`) L7205 - **§3-Two** — `reconcile_tu` bugs found underneath, one introduced while fixing the other L7228 - **§100** — Prefer the DRAFT-LOCAL fix: a type only one function uses belongs in its BODY, not in a shared header (Phase 29 SESSION-22, `func_80175DA8` 0/137 → 137/137) L7254 - **§101** — The STALE DEFAULT class: a guard whose cause was removed is a silent skip wearing a safety label (Phase 29 SESSION-22, three instances in one session) L7283 - **§102** — A PLUMBING verdict can MASK a DIFF; and K&R is not always a codegen change (Phase 29 SESSION-22, `func_8016EC0C`) L7310 - **§103** — A FILE-scope `extern` in a shared overlay TU is a GLOBAL constraint on every LATER function; move the DECL, not the draft (Phase 29 T48/T51, `func_80135260` — the fleet-wide half) L7345 - **§104** — Two silent-skip defects in one scan: match on MASKED text, emit from the ORIGINAL (Phase 29 T53, `gather_externs`) L7433 - **§105** — A gate's revert must survive an EXCEPTION, not just a failure (Phase 29 T53, `jtbl_family_bank`) L7469 - **§106** — Persist the MEASUREMENT, derive the POLICY: a stored route let a stale file out-vote the live table (Phase 29 T54, `residual_class._ROUTE`) L7508 - **§107** — A lever wired into ONE gate path is a lever most families cannot reach (Phase 29 T56, `func_80144090` 0/136 → 136/136) L7551 - **§108** — Diagnosing a family `0/N`: the four causes, and the third opt-in lever (Phase 29 T59) L7589 - **§109** — Conforming a definition to a shared header: fix the NAMES, then check the RETURN precondition (Phase 29 T60) L7638 - **§110** — A unit must define exactly ONE function, and "ends in `;`" does not tell you which line defines it (Phase 29 T65) L7678 - **§111** — The distinct-code metric is not noisy: a family pays it only if its members are byte-VARIANT (Phase 29 T66) L7719 - **§112** — A macro-scoped declaration only collides where the macro is INSTANTIATED (Phase 29 T67/T69, `audit_header_sigs.py`) L7768 - **§113** — An ARITY blocker only exists if the macro CALLS the function; an address-taken use has no call site (Phase 29 T72) L7814 - **§114** — The THIRD decl axis: a CALLEE the draft declares differently from the target TU (Phase 29 T76/T77) L7846 - **§115** — A `func_XXXXXXXX` predicate rots by design: the same name-form assumption in THREE places (Phase 29 T78) L7888 - **§116** — Optimization level is a property of the FILE, not the function: read a family 0/N against the member's stub HOME (Phase 29 T79) L7915 - **§3-The** — fix moves the DEFINITION, not the stub — and here is why the obvious shortcut fails L7933 - **§117** — Spell the sibling's symbol from the SIBLING's address, not the exemplar's kind (Phase 29 T82) L7964 - **§3-Why** — it survived so long: a MASKED oracle will MATCH a wrong symbol L7991 - **§118** — Ordinal (positional) immediate resolution: compare C tokens to the DIFFERING asm uses (Phase 29 T87) L8001 - **§119** — Two levers on the SAME axis, opposite directions: test the off-diagonal (Phase 29 T89) L8035 - **§120** — Uniquify draft-defined TYPE names; and check which of N staging sites you actually patched (Phase 29 T93) L8064 - **§3-Do** — NOT "strip the duplicate typedef" — it breaks the extern that uses it L8075 - **§3-The** — wiring trap that cost two attempts L8081 - **§121** — Synthesise externs for macro-DEFINED callees from the macro's own definition head (Phase 29 T95) L8099 - **§122** — GATE RAW BEFORE TRANSFORMING; the undo belongs to the WRITER, as a per-edit journal (P30 T0a, 2026-07-30) L8122 - **§123** — PROPAGATE A FAMILY WITH THE TOOL ITS TIER NEEDS: `dedup_propagate` is h_exact-only; its refusals are statements about the TOOL (P30 wave 1, 2026-07-30) L8154 - **§124** — A "not matched" verdict can mean the definition is there under a DIFFERENT C NAME: the asm-label alias blind spot (P30 SESSION-28, `func_8016191C` ×137) L8191 - **§124a** — a family sweep's `0 matched-exemplar families` may be a FILTER, not a wall L8236 - **§125** — Split the CARVE from the BODY before calling a jr residue a wall — and measure it by SHA from a CLEAN tree (P30 SESSION-28; **this section's first draft was WRONG and the method caught it**) L8244 - **§3-The** — method (keep this) L8251 - **§3-The** — instrument rules that make its answer trustworthy (this is where I failed) L8261 - **§3-The** — corrected results (each SHA-verified, from a clean tree, restore re-verified) L8276 - **§3-Two** — further notes worth keeping L8286 - **§3-The** — meta-lesson L8295 - **§126** — The carve-within-a-carve: an ADDRESS RANGE is not an OPTIMIZATION REGION (P30 T2, byte-proven end-to-end) L8302 - **§3-The** — finding: opt level is per FILE, so the file's contents must be opt-HOMOGENEOUS L8317 - **§3-The** — instrument trap that hid it (and it is §124's shape again) L8328 - **§3-The** — mechanics L8337 - **§126a** — a bare `except: continue` around a coverage-asserting oracle re-creates the silent skip (P30 S28) L8355 - **§127** — The `-O0` regime: the CONSTANT-OFFSET FOLD, and why `-O0` needs its own idiom set (P30 T3 wave, 15 targets) L8387 - **§3-The** — idiom they kept re-deriving: the constant-offset fold L8394 - **§3-The** — rest of the `-O0` regime (write PLAIN C, and mean it) L8405 - **§127a** — §71 (sibling-first) is the strongest `-O0` lever, and it beats the index L8415 - **§127b** — the knowledge was in a SOURCE COMMENT, not the cookbook L8424 - **§128** — A raw NUL in C source makes grep SILENTLY SKIP the file (P30 S28, 137 files) L8431 - **§128a** — a negative control must corrupt a SCRATCH COPY, never the tracked file L8463 - **§129** — Post-carve, `rtu_match`/`match_one` COUNT THE JUMP TABLE AS INSTRUCTIONS; and a carve must never be committed without its owner (P30 S28, `func_8013BD74`) L8475 - **§129a** — the target instruction count is INFLATED after a carve L8479 - **§129b** — never commit a carve whose owner is still a stub (it strands the carve) L8500 - **§3-The** — real blocker underneath, for the record L8515 - **§130** — An INCREMENTAL build can report BYTE-IDENTICAL for a change the CLEAN build cannot even LINK (P30 S28, the jr pair) L8524 - **§3-The** — diagnostic ladder that finally located it (reusable) L8561 - **§131** — The jtbl OVER-SPAN: `sltiu N` is ground truth in BOTH directions, and the zero-word rule only guards one (P30 S28, `func_80191C50`) L8571 - **§132** — The `JR-PAIR-IN-ONE-O0-OBJECT` "wall" was TWO instrument defects: a merged-double span the carve could not see, and a truncated object no rule deleted (P30 S29, `func_8013B83C` + `func_8013BD74`) L8608 - **Defect** — 1 — a pre-§8e MERGED DOUBLE is not a single-table predecessor L8617 - **Defect** — 2 — `as` writes a corpse and nothing deletes it L8635 - **§3-The** — fingerprint, and the 30-second ladder that found it L8645 - **§3-The** — transferable rule L8666 - **§132a** — `--like` is for a sibling with NO record; against one that HAS a record it over-derives (P30 S29, `ov_SC07_010`) L8673 - **§132b** — When the span's already-matched owner is ITSELF multi-switch: `--span-rel` (P30 S1, `func_8014032C`) L8698 - **§133** — The DEFAULT-FILTER class: three times in one session, a tool silently answered a narrower question than the one asked (P30 S1–S3) L8723 - **§134** — MULTI-LINE BLINDNESS: one root cause, four faces, in `family_remap`'s preamble scanner (P30 S6, 190 zero-crack families) L8749 - **§135** — Six byte-verified gcc-2.7.2 idioms from the P30 S6f-h waves (and the two-lane wave shape that found them) L8803 - **§3-The** — codegen idioms L8808 - **§3-The** — integration idioms (these decide whether a byte-correct draft BANKS) L8842 - **§3-The** — wave shape that produced these L8857 - **§136** — The LOCAL-VARIABLE lever: how many C locals, at what scope (P30 wave 4a, 25 byte-verified banks) L8875 - **§3-The** — splitting/merging rules (each closed a residual, byte-gated) L8891 - **§3-The** — type-form rules L8917 - **§3-The** — scheduling rules (refining §135-2 and §135-4) L8950 - **§3-The** — declaration surface (integration, not codegen) L8979 - **Wave** — economics (measured, for the next batch's sizing) L8991 - **§136a** — Blocker capture: classify on the OUTPUT, never on the exit status L9014 - **§136b** — A prior wave's "genuine byte-DIFF" verdict is NOT reliable evidence (4 of 4 refuted) L9071 - **§136c** — SIBLING-FIRST is a DERIVATION shortcut, not just a conflict fix (the fastest route in a family wave) L9107 - **§136d** — Four gcc-2.7.2 levers the redraft lane found (each closed a residual no other lever moved) L9132 - **§136e** — §136c's PRECONDITION, and two more symptom keys (wave 4b batch 3) L9176 - **§136f** — Two declaration sub-cases the reconcile lane surfaced (lane now 15/15 lifetime) L9224 - **§136g** — When the index points at the WRONG lever: two byte-refuted routings (func_801863B4) L9249 - **§136h** — CORRECTION: the zero-crack pool does NOT "refill with cheap work" (my error, byte-measured) L9278 - **§136i** — The drafter model LADDER: Haiku → **Sonnet** → Opus → Fable5 (Drew, 2026-08-03) L9308 - **§136j** — The failure MIX flips with function size (measured across four bands, one session) L9340 - **§137** — REGALLOC-PERM is a TWO-COMPILE ARITHMETIC PROBLEM, not a permuter job L9382 - **§137a** — A gate verdict has a TIMESTAMP; re-check it against the draft's mtime L9424 - **§138** — The propagation lanes: a gate refusal is a DECLARATION, and which lever you owe depends on blast radius L9453 - **§3-The** — triage, cheapest first L9459 - **§3-The** — DEFINITION-side alias is the only escape when the fleet canon disagrees on a promoting param L9511 - **Rank** — the lane by measured concentration, not by class count L9519 - **THREE** — carry variants hide in one "CARRY-FIXABLE" bucket — and they need different fixes L9528 - **§134** — again, in a second tool — and the waiter rule corrected L9564 - **STEP** — 0 of sibling-first: grep `src/` for a distinctive LITERAL from the `.s` L9584 - **Reconciling** — a gate-refused draft: which way you edit depends on WHERE the TU's decl is L9603 - **§139** — A GATE THAT GREPS FOR VERDICTS MUST ASSERT 1:1 ACCOUNTING; and a `--src` filter must not survive a carve (P30 S38, wave 6: 10 of 16 drafts vanished) L9632 - **§3-The** — generalisation — three corollaries worth more than the bug L9664 - **§3-And** — the inverse-lookup trap, same session L9681 - **§140** — A METRIC IS NOT A MEASUREMENT UNTIL IT IS REPRODUCIBLE FROM THE COMMITTED TREE (P30 S1e: a phantom regression that gated the session's best lever) L9698 - **§3-The** — three-line proof (do this before diagnosing any metric movement) L9718 - **§3-The** — two instrument defects it exposed L9727 - **§3-The** — same swallow, twice more, in the integration spine L9747 - **§3-Two** — wrong mechanisms I chased first, and why they were wrong L9758 - **§141** — The §134 class is CLOSED: every line-shape decision now routes through `cdecl._mask` (P30 S39) L9778 - **§142** — An open stub whose `h_exact` class is MATCHED elsewhere is FREE. Propagate the body; do not gate a draft. (P30 S39, +7,710 ins in two commands) L9816 - **§3-The** — measurement (do this before any wave; it is ~20 lines and needs no builds) L9828 - **§3-The** — trap that hid it — SAME FUNCTION, TWO ROUTES, ONLY ONE IS FREE L9843 - **Route** — selection (why `--addr` sometimes says "nothing changed") L9852 - **§3-And** — the report-vs-bytes lesson attached to it L9860 - **§143** — `cast_call_sites` read a RETURN STATEMENT as a prototype and deleted it. A 0/39 sweep became 18/39. (P30 S40) L9872 - **§144** — THE LITERAL'S SPELLING PICKS THE IMMEDIATE ENCODING (P30 S40 wave 1, `func_801822E0`) L9926 - **§145** — Three loop/combine levers from the S40 wave-2 drafters (16/16 match_one) L9965 - **§146** — RE-MEASURE A WALL BEFORE YOU RESPECT IT. Both "permanent" giants fell to drafts already on disk. (P30 S6, +50,094 ins) L10015 - **§3-Why** — a correct draft can read as an intrinsic wall L10026 - **Then** — propagation returned 0/137 TWICE — both times a missing TYPE L10038 - **§3-Two** — errors of mine, both instructive L10047 - **§3-The** — rule L10061 - **§147** — The three-stratum FRAME LAW, and four "stop searching" verdicts (P30 S42, `func_8017C294`, serial run) L10074 - **§3-A.** — The frame has THREE strata, and stratum 3 is unreachable from C L10079 - **§3-B.** — A `?:` on MEMORY operands costs ~16 bytes of invisible frame; on REGISTER operands, zero L10095 - **§3-C.** — Inner-block declaration does NOT delay slot allocation — BYTE-REFUTED L10106 - **§3-D.** — A lone `$t8`/`$t9` in the target is RELOAD SCRATCH — reproduce the spill, don't pin the register L10111 - **§3-E.** — A `qty_compare` TIE is not spelling-reachable — recognise it and stop L10121 - **Consequence** — for the family (a real scheduling decision) L10133 - **§148** — The loop.c hoisting THRESHOLD is arithmetic you can compute, and the `?:` clamp that folds to MIN_EXPR (P30 S42, `func_8017C6F4`, 947 ins) L10183 - **§3-A.** — `move_movables` hoists iff `threshold × savings × lifetime ≥ insn_count` — and you can read it L10189 - **§3-B.** — `(v < 0x40) ? v : 0x3F` is folded to `MIN_EXPR` and expands to the WRONG SHAPE L10208 - **§3-C.** — A zero-byte ALLOCNO-PRIORITY slider L10222 - **§3-D.** — Reproduce the original's BUGS verbatim L10232 - **§149** — Four instrument defects in one session, and the two questions they were hiding (P30 S43) L10281 - **§3-A.** — A prep step that returns its input on failure is indistinguishable from a search that found nothing L10287 - **§3-B.** — Same address + same name ≠ same body — and the ledger keys on address L10305 - **§3-C.** — `make: *** [...] Error N` is a summary, never a diagnosis L10321 - **§3-D.** — "Cheap fuel" that was never probed: 0 of 31 templatable L10331 - **§150** — A register ROTATION across symmetric blocks is VARIABLE-IDENTITY evidence, not an allocator tie (P30 S43, `func_8017C6F4`, 947 ins ×4) L10350 - **§3-The** — fix L10357 - **§3-The** — method that found it (this is the transferable part) L10367 - **§3-Two** — corrections to the record L10383 - **Diagnostic** — order (adopt this) L10394 - **§151** — THE GHOST WEDGE: when a load-before-store transposition is unreachable by ANY statement order (P30 S43, `func_8017EF68`, 969 ins) L10405 - **§3-The** — mechanism (read from cc1's own `-dR` trace, not inferred) L10410 - **§3-The** — lever — a zero-emission insn that absorbs the blocked tick L10425 - **§3-The** — two fallouts, and how to close them (both measured, in order) L10433 - **When** — to reach for it L10445 - **§152** — BYTE SIZE is the family key that name- and h_seq-grouping both miss (P30 S43, the 0xECC family: 1 crack → 12 overlays → 11,364 ins) L10456 - **§3-The** — finding L10461 - **§3-The** — key L10470 - **§3-Two** — cautions that must travel with this technique L10481 - **§3-The** — companion defect (open) L10492 - **§153** — THE ADDRESS-REMATERIALISATION LAUNDER: a third zero-emission asm lever (P30 S43, `func_8018D98C`, 710 ins) L10504 - **Symptom** — Symptom L10512 - **Mechanism** — (gcc source + RTL dumps, not inferred) L10517 - **What** — does NOT work (14 byte-measured probes) L10524 - **§3-The** — cure — a fresh launder per site, each in its own block L10530 - **Companion** — levers from the same function L10540 - **§154** — Reading a disc payload: the module-id word, static base derivation, and "type 1 = uncompressed overlay" (P30 S44) L10560 - **§3-A.** — Payload word0 is a global MODULE ID; code starts after the header L10565 - **§3-B.** — Two static base-derivation methods that must AGREE (use both) L10574 - **§3-C.** — PAC type 1 = the same payload class as type 4, just NOT compressed L10586 - **§155** — hi/lo literal scanning MUST track base registers (S45) L10596 - **§155a** — the same failure class, one level up: SHAPE-blind table scanning (S45 p5) L10605 - **§155b** — check the TYPE your oracle returns before comparing against it (S45 p5) L10625 - **§155c** — the ZERO-REFERENCE trap: gcc splits a global-array address across the `lui` and the LOAD (S46) L10648 - **§156** — an ORPHANED reconcile poisons the fleet: `dedup_propagate`'s kept edit (S45 p6/p7) L10678 - **§157** — the cheap-tier size cliff, measured (S45 p6) L10725 - **§158** — The RANGE-EXTENDER: a fourth zero-emission asm lever completes the allocno toolkit (P30 S46 tier-3, `func_8017CE58`, 733 ins) L10750 - **Symptom** — Symptom L10759 - **§3-Why** — the fork is chained (gcc source, validated insn-by-insn against -dS/-dR dumps) L10765 - **§3-The** — method (dump-arithmetic first, then place — no probing) L10778 - **Bonus** — facts worth keeping L10795 - **§156** — THE PREFERENCE-DONOR MERGE: cross-region variable reuse is what fills a0-a3, and a call-arg use in ONE region steers the fill in ALL of them (P30 S46 tier-3, `func_80186E24`, 611 ins: 236-off "S11 regalloc-order" → MATCH, zero new pins) L10810 - **§159** — THE DECLARATION AXIS: conform to byte-truth, and make every guard state its COVERAGE (P30 S47; ~10,930 sites across 8 axes, fleet byte-identical) L10866 - **§160** — THE REACH-15 WAVE HARVEST: an align-1 block move, and the instrument that called it a wall (P30 S47) L10931 - **§161** — THE RETRY-WAVE HARVEST: a jump table indexed from zero, and two allocator traps (P30 S47) L11007 - **§162** — S48 WAVE-1 HARVEST (P30, 2026-08-11): the reach-ordered sibling campaign's first 12 targets L11052 - **§162a** — SHARPENS *(sharpens §161a, §131, §8a-pad, §129a)* L11079 - **§162b** — SHARPENS *(sharpens §48-A3, §156, §150, §76)* L11108 - **§162d** — SHARPENS *(sharpens §31, §21, §30, §55a)* L11137 - **§162e** — NEW L11179 - **§162** — THE LICM PAIR: what makes an address a movable AT ALL, and why the preheader order is the body order (P30 S47, `ov_MAIN_012`) L11181 - **§162f** — SHARPENS *(sharpens §42d, §41d, §73, §10)* L11247 - **§162g** — NEW L11302 - **§162** — CROSS-JUMP DIRECTION: the surviving copy is always the LATER one, so a BACKWARD `j` into a sibling arm is a source `goto` (P30 S48) L11304 - **§162h** — SHARPENS *(sharpens §88, §88a, §50-B, §8)* L11338 - **§162** — The cross-jump "CALL veto" is a COUNT law, not a CALL law (BOUNDS §88a; P30 S48, `func_80189540`) L11340 - **§162i** — SHARPENS *(sharpens §135, §21, §42, §32)* L11406 - **§162j** — SHARPENS *(sharpens §25, §136d-1, §48-B, §46-L2)* L11431 - **§162k** — SHARPENS *(sharpens §1-I2, §12, §160d, §21)* L11460 - **§162l** — SHARPENS *(sharpens §48-B, §48-C1, §20, §21)* L11523 - **§162m** — SHARPENS *(sharpens §36, §158, §148, §153)* L11578 - **§158a** — THE FIFTH LEVER IS NOT AN ASM: `do { } while (0)` is a REGION ref-multiplier you MINT (P30 S48, `func_8017CBC8`, ov_MAIN_012 / jr_801789AC, 188 ins → MATCH) L11580 - **§3-The** — law L11587 - **Size** — it before you write it (§158 step 1-2, applied) L11598 - **§3-The** — wrap BOUNDARY is the dial — and it is indiscriminate L11619 - **§3-Not** — a pure dial L11625 - **DIAGNOSTIC** — TELL — two faces, one law L11629 - **§162n** — NEW L11651 - **§162o** — SHARPENS *(sharpens §158, §136-1, §136-6, §79)* L11684 - **§162p** — SHARPENS *(sharpens §48-B, §46-L2, §156, §136d-1)* L11743 - **§162q** — SHARPENS *(sharpens §30, §30a, §135-2, §136-13)* L11760 - **§163** — S48 WAVES 2-3 HARVEST (P30, 2026-08-11/12): the five that were byte-probed and are actionable L11798 - **§163z** — THE UNVETTED REMAINDER (do not cite as law; each needs a dedupe pass) L11866 - **§164** — S48 §163z SKEPTIC PASS (P30, 2026-08-12): 190 claims vetted, 82 banked L11886 - **§16Xy** — SHARPENS *(sharpens §136d-3, §37 /s-DEP LATTICE, §135-2, §136-13, §162q)* L12338 - **§3-The** — `/s` drop clause is in ALL THREE dependence predicates, so the FIXED-ADDRESS STORE is what floats (P30 S48, `func_8017CA18`, ov_MAIN_012) L12340 - **§164z** — REFUTED CLAIMS: do NOT re-derive these L13674 - **§165** — S48 WAVE-4 HARVEST (P30, 2026-08-12): banked the same day the wave landed L13740 - **§16Z** — SHARPENS *(sharpens §37 "the /s-DEP LATTICE", §136-13, §136-14, §16Xy, §162q, `gcc-2.7.2-map/sched.md` §64)* L14398 - **§3-The** — ADDRESS-CLASS TABLE: which load/store pairs even REACH the `/s` clause (P30 S48 wave 4, `func_80185B44`, ov_SC03_014) L14400 - **§165z** — REFUTED THIS WAVE: do NOT re-derive L14916 - **§166** — THE DESTINATION-TU ORACLE (P30 S48): the seven-attempt bug that was never codegen L14960 - **§167** — S48 WAVE-5/6 HARVEST (P30, 2026-08-12): the saturation point L15016 - **§167z** — REFUTED IN WAVES 5/6: do NOT re-derive L16212 - **§168** — THE COUSIN TIER (P30 S49, 2026-08-12): h_seq's exact-hash brittleness, measured — and the similarity map above it L16269 - **§169** — THE MICRO-ADAPT LANE (P30 S49): edit a proven body, don't crack a new one L16321 - **§170** — THE A-PROP WORD-DIFF CARD (P30 S49): the lane that had no owner L16368 - **§171** — THE STALE SEED SYMBOL (P30 S50, 2026-08-13): why §170's 91%→57% was never codegen L16412 - **§171a** — THE MECHANICAL A-PROP DRAFT (P30 S50): 256 members banked with no agent in the loop L16478 - **§171b** — THREE CARRIES THE MECHANICAL DRAFT NEEDS (P30 S50, banking the top-reach families) L16528 - **§172** — THE ORPHAN-SLOT MECHANISM v2 (P30 S50-Max): the complete frame-residue model for gcc-2.7.2 MIPS L16561 - **§172a** — TWO DECOMPILATION TELLS FROM THE SAME DIG (P30 S50-Max) L16615 - **§172b** — THREE MORE TELLS FROM THE GCC READ (P30 S50-Max, banked on Drew's ask) L16631 - **§173** — THE STORED-PLUMBING RECOVERY RECIPE (P31 T6): symfix-first, per-group isolation, and where the verdicts have no drafts L16674 - **§174** — THE ADAPT-CARD WAVE RECIPE (P31 waves A/B, 2026-08-14): prevention beats recovery L16710 - **§175** — A CALLER-SAVED REGISTER PIN CAN BE A CORRECTNESS BUG, NOT JUST A SCHEDULING CHOICE (P31 wave H, 2026-08-15) L16780 - **§176a** — THE VERIFICATION-LAYER LAWS (P31 overnight, 2026-08-15). What each check can and cannot prove. L16805 - **§176b** — BATCH-GATING MECHANICS (P31): what changes when N drafts land in ONE .c L16845 - **§176d** — THE CONFLICT TABLE MUST BE SEEDED FROM THE TU, AND KEYED PER FILE (P31 S52, 2026-08-15) L16871 - **§176e** — SYMBOL IDENTITY IS COMPUTABLE OFFLINE (P31 S52): `tools/reloc_identity.py` L16910 - **§176f** — THE DECLARATION FORM IS A MATCHING LEVER, SO RECONCILE TOWARD THE FORM THE MATCH NEEDS (P31 S52) L16967 - **§176g** — SIZE A WAVE BY INSTRUCTIONS, NOT BY CARDS (P31 S52 — the adopted doctrine) L16998 - **§176h** — THE BATCH-SUBSTITUTION HAZARD MAP (P31 S52): seven holes, three wrong fixes, one law L17036 - **§3-A.** — The seven under-reporting holes (all in `gate_main`, all the same shape) L17044 - **§3-B.** — Typedef handling — the only strategy that survives contact L17062 - **§3-C.** — The limit that remains (recorded, not solved) L17084 - **§3-C2.** — RECONCILE BEFORE THE FIRST GATE — a parked draft gets HARDER to bank, not easier L17091 - **§3-D.** — The measured cost shape, and what to build next L17114 - **§176i** — WHAT A STATIC PRE-GATE CHECK CAN AND CANNOT PROVE (P31 S52, wave Q) L17132 - **§176j** — STOPPING A WAVE MID-FLIGHT COSTS THE IN-FLIGHT TAIL (and how much is recoverable) L17164 - **§176j-2** — THE REPAIR PASS, MEASURED (do this instead of resuming) L17190 - **§176k** — TWO SELECTOR BUGS THAT SILENTLY SHRINK A WAVE L17210 - **§177** — 🔴 THE EPILOGUE RETURN-DELAY SLOT IS DECIDED BY YOUR SAVED-REGISTER SET, NOT BY SCHEDULING L17227 - **§178** — SIX LEVERS MINED FROM THE WAVE-P JOURNALS (P31 S52), each byte-proven and source-cited L17284 - **§3-A.** — THE `$0`-ADD OPAQUE COPY defeats `make_regs_eqv` (func_80033398, 93 ins) L17296 - **§3-B.** — A `return ` IS A PRIORITY-1 HARD-REG SET THE SCHEDULER PLACES FIRST (func_8001BE30, 92 ins) L17312 - **§3-C.** — SINGLE-SET TEMPS GET THE BIRTHING BOOST (func_8001D3FC, 196 ins) L17325 - **§3-D.** — A NARROW TYPE BLOCKS COPY ELISION (func_8001D3FC — new idiom) L17333 - **§3-E.** — THE ZERO-OFFSET ALIAS HOLE (func_80037028, 71 ins) L17340 - **§3-F.** — `MEM_IN_STRUCT_P` ASYMMETRY IN `true_dependence` (func_80037144, 124 ins) L17349 - **§3-G.** — TWO MODELLING TRAPS THAT COST THESE AGENTS SWEEPS OF HUNDREDS OF COMPILES L17355 - **§179** — IDIOMS MINED FROM THE WAVE P/Q JOURNALS (P31 S52, harvest pass) L17371 - **§179-A** — 🔴 A LOOP-WALKED POINTER **PARAMETER** HANDS ITS ARGUMENT REGISTER TO THE GIV (9 byte-proofs) L17381 - **§179-B** — 🔴 THE HAND-WRITTEN-ASM TRANSCRIPTION CHECKLIST (maspsx + masked_diff; 10 byte-proofs) L17430 - **§179-C** — 🔴 A FUNCTION WITH NO EPILOGUE (falls into a sibling's shared tail) MUST BE FILE-SCOPE `__asm__` L17480 - **§179-D** — `gte_stflg` MUST CLOBBER `"$12"` OR THE WHOLE TEMP FILE ROTATES BY ONE L17531 - **§179-E** — A `>2*MAX_MOVE_BYTES` BLOCK COPY IS A **STRUCT ASSIGNMENT**, NOT A HAND LOOP L17551 - **§179-F** — PINNING A LOOP-WALKED POINTER IS A TOTAL OFF-SWITCH FOR STRENGTH REDUCTION L17577 - **§179-G** — 🟡 A PIN CAN **CREATE** A COMBINE `LOG_LINK` AND DELETE AN `andi` (sixth RC-5 channel, n=1) L17600 - **§179-H** — A MID-BODY `.global LABEL` PAIR SLICES A BYTE-COMPARABLE FRAGMENT OUT OF A LARGER ROUTINE L17627 - **Considered** — and NOT banked L17650 - **§176c** — MAIN (SLUS_007.26) CANNOT BE GATED INCREMENTALLY L17667 - **§176** — SEVEN LEVERS FROM THE P31 OVERNIGHT WAVES (2026-08-15): statement order, false regalloc, and the pin that fights back L17678 - **§176-A** — "SCHEDULE / DELAY-SLOT / LENGTH-DRIFT ±1" ⇒ check STATEMENT ORDER around the call first L17684 - **§176-B** — "REGALLOC-PERM, 1-4 instructions off" ⇒ it is usually NOT register allocation L17708 - **§176-C** — 🔴 WALL REFUTATION: a hard-register pin CANNOT schedule around a call, because of a genuine gcc-2.7.2 bug L17754 - **§176-D** — CSE-class levers used in reverse (two sharpenings of §153 and cse_expr §2) L17780 - **§176-E** — Two cheap source spellings, both cc1-probed L17806 - **§176-F** — Misdiagnosis triage: four residual verdicts that were lying L17828 - **What** — is NOT banked here L17845 - **§180** — THE LEFTOVER-DRAFT HARVEST: RE-VERIFY WHAT YOU ALREADY HAVE BEFORE DRAFTING ANYTHING NEW L17858 - **§180b** — WHAT THE PRE-GATE LADDER ACTUALLY FINDS IN A COLD PILE (the shape of integration debt) L17887 - **§180c** — WHEN A BINARY'S MASS BAND IS SPENT, THE FLEET-WIDE DRAW IS STRICTLY BETTER L17906 - **§181** — WHAT A WAVE'S GATE ACTUALLY REJECTS (P31 S53, measured on wave R's 45-draft main pile) L17924 - **Only** — ONE of 27 blocked drafts was wrong. The other 26 were correct and unbankable. L17925 - **§182** — §177's HONEST NEGATIVE: the epilogue lever cracked 4 of 16, and the `800c3` cluster held L17980 - **§180d** — THE `pgrep` BRACKET TRICK PROTECTS THE PATTERN, NOT THE COMMAND LINE L17994 - **§183** — THE DECLARATION-RECONCILIATION PLAYBOOK (P31 S53, measured on 20 byte-verified drafts) L18006 - **§3-18** — of 20 reconciled while keeping the match. The two that did not are mechanism, not effort. L18007 - **§184** — COMMENT-BLINDNESS IS A DEFECT CLASS, NOT A BUG (P31 S53: three tools, one root cause, one session) L18077 - **§185** — EDIT THE SIDE THAT IS CHEAP TO VERIFY, AND CHECK A TU RETYPE AT ITS USE SITES L18105 - **§186** — CROSS-JUMPING RUNS **AFTER** SCHEDULING, SO NO C-LEVEL BARRIER CAN STEER IT L18137 - **§186b** — A NO-SAVE 16-BYTE FRAME IN A LEAF FUNCTION MEANS `s16` LOCALS, NOT A HIDDEN CALL L18158 - **§186c** — WHERE A VALUE IS LOADED DECIDES WHICH ALLOCATOR OWNS IT, AND THEREFORE ITS REGISTER L18166 - **§187** — 🔴 "SAME SOURCE" IS NOT "SAME OBJECT": THE SDK BUILD AND THE GAME BUILD DISAGREE ON GTE NOPS L18183 - **§188** — 🔴 THE `jr $ra` + `addiu $sp` TAIL IS AN **ASSEMBLER** ARTIFACT, NOT A FRAME SHAPE L18221 - **§189** — FIVE COMPILER LAWS MINED FROM THE WAVE R/S JOURNALS (P31 S53), each source-cited and re-derived by a second agent L18271 - **§190** — THREE PRESCRIPTIONS FROM THE SAME HARVEST (weaker evidence than §189, honestly labelled) L18346 - **§191** — WHAT THIS HARVEST DID **NOT** BANK (4 rejected, 4 narrowed) — recorded so it is not re-derived L18383 - **§192** — THE PRE-GATE LADDER WAS MAIN-ONLY, AND NOBODY COULD SEE IT (P31 S54) L18405 - **Three** — defects in one call path; the overlay slates that carry most of the wave work were being waved through L18406 - **§193** — THE WAVE-T HARVEST (P31 S54): 71 index_gap reports -> 9 laws, 5 rejected, 61 already-covered L18470 - **§193-A** — The wave card ships only OPEN-set pointers and discards the atlas's BANKED one — `exemplar`/`sibs` are stubs 100% by construction (tools/atlas.py:96/657), while `seed.ref` (matched pool, atlas.py:505-536) is dropped at build_wave_atlas.py:143 L18487 - **§193-B** — A NARROW CAST OF A WIDE PARAMETER COSTS ONE EXTRA INSTRUCTION WHEN ITS STATEMENT SITS AFTER AN INTERVENING `jal` — AND THE DECIDER IS `combine.c:929`'s CROSS-CALL GUARD, NOT REGISTER ALLOCATION L18554 - **§193-C** — gcc-2.7.2 cross_jump merges the SCHEDULED common SUFFIX only — there is no prefix/head merge, so §8/§48-A1's "duplicate into both arms and cross_jump refunds it" is a TAIL-only lever L18592 - **§193-D** — A COMPILER-GENERATED ARGUMENT COPY IS A `optimize_reg_copy_1` TRIGGER: when the pointer's LAST use in a block precedes a call that takes the pointer *as a register*, sched1 hoists the implicit `move $aN,$sN` to the block top and local-alloc re-bases the WHOLE block's memory operands onto `$aN`. The only C dial is a label (the §165-24 goto-join) between the block and the call. L18626 - **§193-E** — A varying-address (pointer) load is re-emitted once per CSE-LIVE INTERVAL, and naming it in a C local is the only C-level lever over that count — no store SPELLING has any reach L18697 - **§193-F** — §148-A2 — The `threshold -= 3` STAIRCASE: `move_movables` admits a COUNT of invariants, not a boolean — two identical merged constants split hoisted/not-hoisted by LIST ORDER, and `insn_count` picks the rank cutoff L18727 - **§193-G** — §164-54's "scope to ≥4 arms" bound is byte-wrong — the dispatch-topology oracle goes live at THREE case nodes (`balance_case_nodes` splits at `i > 2`), but only for a signed-after-promotion index L18791 - **§193-H** — A pointer-derived base load `*(s32*)(p+K)` is uncacheable across ANY memory write or call (cse's third kill disjunct) — so the target emits one `lw` per store-separated RUN of spellings, and a run of stores sharing one `lw` is a C local you must declare L18852 - **§193-I** — A DECLARED AGGREGATE LOCAL HAS AN 8-BYTE FRAME STRIDE — CEIL(size,8), NOT size. N ARRAYS THEREFORE COST 8·k MORE THAN ONE HAND-LAID STRUCT (k = how many of them have size mod 8 ≠ 0), AND THE TELL IS IDENTICAL INSTRUCTION COUNT WITH EVERY $sp DISPLACEMENT SHIFTED BY THE SAME CONSTANT. L18892 - **§193-REJECTED** — what this harvest did NOT bank (recorded so it is not re-derived) L18951 - **§194** — THE WAVE-U HARVEST (P31 S54): 64 index_gap reports -> 14 laws, 5 rejected, 44 already-covered L18978 - **§194-A** — A zero-byte scheduling fence goes AFTER the defining statement to make that computation emit FIRST in its block — and the barrier predicate is `volatile`-or-colon-less, not the `"memory"` clobber (COMPLEMENTS §165-40; does not refute it) L18995 - **§194-B** — A `addu $rA,$rB,$zero` copy feeding ≥2 `sh` stores is a SECOND, 16-BIT-DECLARED local — the width, not the clamp or the join liveness, is what keeps the copy (and func_80183094's pin + `"memory"` clobber are both provably inert) L19061 - **§194-C** — A CALLER-SAVED loop counter proves its live range crosses ZERO calls — so it cannot share a pseudo with any value LIVE ACROSS a call (but it may freely share one with values that merely sit between calls) L19099 - **§194-D** — Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication + attribution correction of §165-17, NOT a new argument-register law L19162 - **§194-E** — The wave card's `exemplar` is the TARGET ITSELF on 42/73 wave-U and 36/71 wave-T cards — a construction consequence of `atlas.py:657` (exemplar = max-nins OPEN member) meeting `build_wave_atlas.py:166` (one card per gid); and the shipped §193-A `seed_ref` is same-binary on 0/51, so no card field can ever name a destination-TU sibling L19203 - **§194-F** — `if ((*p = v = f()) == 0)` is an expand-time pseudo SPLITTER (store_expr's `want_value && MEM` path), not a fold — it partitions one value between local_alloc and global_alloc. BOUNDS §21's L1872 bullet, whose stated direction is byte-wrong on 3 of 4 instances, and CLOSES the control §167-37 asked for. L19240 - **§194-G** — Reading the integer half of a 16.16 stack aggregate: REGISTER-LIVENESS, not the C spelling, decides `sra` vs `lhu` — and the break is count-neutral L19313 - **§194-H** — §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a branch fences the store that reads v, and when TWO independent stores compete for the one delay slot the fence picks the winner at ZERO length drift (a WIDTH/sh!=sw swap, not a nop). Corrects §167-42's reconstruction (pseudo arm → hard-reg arm) and supplies its first re-runnable A/B. L19356 - **§194-I** — §16N+2's magic-per-odd-part ladder has exactly one broken row — read the divisor arithmetically instead: d = round(2^(32 + post_shift) / magic_read_as_unsigned) L19394 - **§194-J** — Back-to-back identical stores: flow.c's `last_mem_set` deletes the first, and only `volatile` saves it L19437 - **§194-K** — Blind sched1's alias oracle with a second SET of a pointer pseudo — the first zero-byte, non-volatile, dependence-CREATING lever (corrects §167-05's "volatile is the only door"; fourth consumer of reg_n_sets) L19489 - **§194-L** — §88b and §189-E are BOTH half-wrong, but not the way the candidate says: the compare-constant shape is a 2-D lookup (cmp_info ROW × constant-in-window), and naming matters on OPPOSITE sides of the window for the GE/LT rows vs the GT/LE rows L19567 - **§194-M** — A STORE in a CONDITIONAL branch's delay slot proves its C statement DOMINATES the branch — reorg can never pull a store out of either thread (gcc-2.7.2, -mips1) L19615 - **§194-N** — §193-D's C dial is misstated: the lever is a SURVIVING CODE_LABEL (a label with a real incoming edge), not "a label between the block and the call" — a bare label, or a `goto L; L:` pair whose target is the next active insn, is deleted by jump1 (jump.c:663-669 → delete_insn → jump.c:3458-3461, and jump.c:243 for the bare case) long before sched1/local-alloc, and costs exactly zero bytes L19673 - **§194-REJECTED** — what this harvest did NOT bank (recorded so it is not re-derived) L19733 - **§195** — THE WAVE-V HARVEST (P31 S54): 67 index_gap reports -> 14 laws, 9 rejected, 76 already-covered L19754 - **§195-A** — §167-08's "an $aN READ before the jal is scratch" has a byte-proven FALSE-NEGATIVE class: an argument that DIES at the call is allocated straight into $aN, so its only def is a plain load far above the jal and every intervening use reads $aN — there is no positive tell in either direction, only the two-arity A/B L19768 - **§195-B** — A CALL_INSN does not start a basic block in gcc-2.7.2 — so a call-crossing temp can be a LOCAL-alloc quantity (the missing precondition under §48-A2 / §52 / regalloc.md K8) L19839 - **§195-C** — A call-argument `%hi/%lo` pair sitting at the block head, far above its `jal`, is a load-delay-gap filler chosen by SOURCE STATEMENT ORDER — swap the two independent statements nearest the call; a single-use `void *p = &SYM;` call-arg temp is OUTPUT-inert against it (but NOT expand-stream-inert) L19881 - **§195-D** — §195 — `masked_diff.mask_for` returns 0 for EVERY `j`/`jal` word, so an internal `j` destination is invisible to match_one, the permuter scorer AND every similarity tier: a control-flow semantic error (which calls execute) surfaces as a 1-instruction residual mislabelled `DELAY-SLOT / profile=schedule`, or as an outright false MATCH L19929 - **§195-E** — A 0/1 materialised at a JOIN immediately before the controlling `beqz`/`bnez` proves the source NAMED the condition — a truth expression in an `if`'s controlling position reaches `do_jump`, which has no value path L19997 - **§195-F** — fold-const.c:4825 canonicalises a `?:` whose THEN arm is zero (or constant against a non-constant ELSE) by inverting the condition and swapping the arms — so a ternary's written arm order is byte-inert there, and `c ? 0 : X` is unspellable: it always compiles as `!c ? X : 0` L20055 - **§195-G** — §NEW — TWO ARMS CALLING THE SAME CALLEE MERGE INTO ONE `jal` UNLESS EACH ARM'S OWN CODE CONSUMES THE RESULT: the C dial that keeps two call sites apart is WHERE the consumer test lives, and the branch SENSE you spell it with is byte-inert (jump.c:1737 canonicalises it) L20107 - **§195-H** — §165-27g — ON A FIXED-SYMBOL GLOBAL, `extern T D[];` + `D[0]` IS THE CSE RELOAD DIAL, AT ZERO ADDRESSING COST, AND THE ELEMENT COUNT IS INERT (§165-27's `T v[2]` CAVEAT IS A LOCAL-FRAME FACT AND DOES NOT TRANSFER) L20155 - **§195-I** — §145(b) AMENDED — the pointer-bump `addiu` is saved by cse, not combine: ANY second SET of a pseudo in the address's equivalence chain (in-place `p += K`, §145(b)'s `p = r;` copy, or an asm re-tie) kills the fold; the pass is cse and the gate is `invalidate`'s `reg_tick++` L20202 - **§195-J** — GTE / PsyQ op CALL-vs-INLINE is a PER-SITE SOURCE FACT, not a TU style — both forms coexist in one TU (measured in 2 TUs), and the tell is the target's own opcodes (`lwc2`/`sqr`/`swc2` vs `jal`), not the sibling. Costs -8 ins on func_8018505C. Corollary: the game's inline `sqr` macro emits TWO hazard nops, the SDK's `Square0` body emits ONE — so the inline form is provably not `Square0` inlined (a second instance of §187's SDK-vs-game GTE nop divergence). L20267 - **§195-K** — At -O2 the §18 array-of-struct lever is a FRAME lever, not a length lever — but only when the N reads carry DISTINCT index expressions; with a SHARED index §18's +2-instruction residual is still alive at -O2 (the submitted "memory-loaded narrow index" precondition and unconditional length-neutrality are both falsified) L20311 - **§195-L** — The cse store-re-seed does not cross a JOIN LABEL: per-arm stores + a join read keep the reload that one join store deletes (bounds §193-E BOUND 1/BOUND 3 with §48-B's EBB boundary) L20363 - **§195-M** — Frame `vars` is a SEQUENTIAL bump-allocation, not a flat sum: §193-I's CEIL(aggregate,8) term and §165-03/§167-06's 8×orphan term are the SAME frame_offset walk at two different compiler stages, and each stage re-CEILs frame_offset to 8 before it allocates L20415 - **§195-N** — In a call-bearing chain of N≥2 `if (f(...)) return 1;` tests closed by `return 0;`, the LAST test must stay in STATEMENT form — the value form (`return f() != 0;` / `? 1 : 0` / `!!f()`) costs +1 `j` and empties the other N−1 delay slots. The cause is REORG block placement, not jump.c's `delete_jump`. L20464 - **§195-REJECTED** — what this harvest did NOT bank (recorded so it is not re-derived) L20511 - **§196** — PUT ON THE CARD WHAT THE TREE ALREADY KNOWS: the fleet's declaration consensus (P31 S54) L20557 - **§197** — THE WAVE-W HARVEST (P31 S54): 68 index_gap reports -> 4 laws, 3 rejected, 41 already-covered L20612 - **§197-A** — A NARROW SIGNED MEMORY READ FEEDING A CONSTANT `>>` LOSES ITS `lh`, AND THE CURE IS AN ASM RE-TIE (attribution CONTESTED: cse vs combine) L20626 - **§197-B** — A REPEATED COMPARE OF ONE VALUE AGAINST ONE CONSTANT IS DELETED BY cse's `qty_comparison_code` CHANNEL (the non-EQ complement of §165-03) — and a front-end-opaque mask on EITHER compare is a pure-C dial that keeps the target's second branch L20670 - **§197-C** — Fix A1 (operand order) cannot move a commutative destination whose .greg conflict set already contains BOTH operand hard registers — split the accumulate so the destination IS the load's pseudo L20698 - **§197-REJECTED** — §197-REJECTED L20743 - **§199** — THE WAVE-X HARVEST (P31 S54/S55): 63 index_gap reports -> 7 laws, 2 rejected, 56 already-covered L20758 - **§199-A** — §189-A's asm→source inference is byte-FALSE: an interloper between a split constant's `lui`/`ori` is a SCHEDULE fact, not a source fact — and the separator is the BIRTHING BOOST, not a "priority floor" (§189-A's split-timing half survives; its "no statement order / no pin" absolute and the candidate's own forward-scheduler narrative both fall) L20769 - **§199-B** — A permutation sweep that holds ANY statement fixed is not a sweep: the statement an agent pins as "obviously load-bearing" is the one carrying the signal, and the partial sweep returns a FLAT residual that reads as proof of order-invariance L20818 - **§199-C** — A NEGATIVE CONSTANT MULTIPLY ALWAYS TAKES expmed's negate_variant — but whether you ever SEE the neg is decided by COMBINE, and for an EVEN |K| it never disappears L20863 - **§199-D** — A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is the WIDTH of a real object, and a widening `(s32)` cast is inert — AMENDS §35 (whose stated "only CSE-reuse canonicalizes" mechanism is byte-wrong) and does NOT apply inside a `switch` L20911 - **§199-E** — §189-A BOUNDED AND CORRECTED — the discriminator is INSN_PRIORITY, not "is the interloper a constant": an insn between a `lui`/`ori` pair proves NOTHING about the source spelling unless it TIES the `ori` on priority, and on the pinned `-mcpu=3000` triple a dependent load never does L20971 - **§199-F** — §164-36b — THE TARGET-HEAD FENCE IS A DELAY-SLOT THREAD SELECTOR, AND IT ONLY FIRES WHEN `mostly_true_jump > 0` (amendment to §164-36; its "−1 instruction" tell is falsified) L21030 - **§199-G** — At TWO case nodes the switch-vs-if oracle is not blind — but the tell is ALL tests positive + a trailing `j default`, NOT the first test's polarity L21084 - **§199-REJECTED** — §199-REJECTED L21139 - **§200** — THE ALIAS IS THE UNIVERSAL DECLARATION ESCAPE: stop negotiating with the TU's spelling (P31 S55) L21150 - **§201** — THE WAVE-Y HARVEST (P31 S55): 67 gap reports -> 5 laws, 8 rejected, 53 already-covered L21199 - **§201-A** — §150-B applies to `decl_prior`'s DEF row: for an overlay-window symbol the banked "definition" is usually another overlay's function, and the card ranks it ABOVE the destination TU L21209 - **§201-B** — In a narrowed PLUS/MINUS/AND/IOR/XOR expression the destination pointee is INERT — the sign of the materialized constant is decided by an OR over the UNWIDENED operands (convert.c trunc1), which bounds §1841 to direct constant stores L21241 - **§201-C** — §X — A CALL'S OWN DELAY SLOT AND THE UPSTREAM CONDITIONAL BRANCH'S SLOT COMPETE FOR ONE INSN (the call's argument copy), AND ONE STATEMENT'S POSITION RELATIVE TO THE CALL DECIDES BOTH — the residual is visible at the BRANCH, not at the call L21321 - **§201-D** — THE SIGNEDNESS OF A div/mod MAGIC IS DECIDED BY THE STATIC TYPE OF THE DIVIDEND TREE AFTER `get_narrower` STRIPS WIDENING CONVERSIONS — NEVER BY A PROVABLE RANGE. AN `& 0xFF` IS NOT A CONVERSION, SO IT NEVER FLIPS THE MAGIC; A DECLARED-UNSIGNED LOCAL *OR* A NARROWING CAST WRITTEN AT THE DIVIDE BOTH DO. L21372 - **§201-E** — §194-J-2 — The `last_mem_set` deletion window is measured in SOURCE/expand order, not in the printed stream; a foreign store moved between the pair in C source is a real lever, and `volatile` is not always the better one L21419 - **§201-REJECTED** — eight, the session's highest L21475 - **§202** — THE ALIAS CARRIES A DEFINITION, NOT JUST A DECLARATION: the DEF-SIDE-RETURN wall (P31 S56) L21500 - **§203** — A DEDUPED TYPEDEF MUST PRECEDE EVERY SPLICE POINT, NOT JUST ITS OWN (P31 S56) L21537 - **§204** — THE WAVE-Z HARVEST (P31 S56): 82 gap reports -> 5 laws, 16 rejected, 30 already-covered L21598 - **§204-A** — A COMPARE THAT APPEARS BOTH IN A BRANCH'S DELAY SLOT AND AGAIN ON THE FALL-THROUGH IS A JOIN WITH TWO INCOMING EDGES: THE TWO GUARDS ARE SEQUENTIAL `if`s, NEVER `if/else if` L21617 - **§204-B** — A LOOP COUNT THAT ARRIVES ON THE STACK IS DECREMENTED IN PLACE: a fresh counter local can cost a real `move` AND permute the whole callee-saved file L21692 - **§204-C** — WHEN A LOCAL BUFFER'S ADDRESS IS PASSED TO A CALL, ITS SIZE IS A FACT ABOUT THE CALLEE'S BODY, NOT ABOUT THE CALL SITE: grep the callee's proven definition and count the stores through the pointer parameter before you declare the local L21756 - **§204-D** — A LOOP-INVARIANT LOAD IS ADMITTED AS A MOVABLE ONLY IF ITS ADDRESS CANNOT TRAP: `local.field` PASSES, THE SAME READ THROUGH A POINTER LOCAL IS REFUSED L21814 - **§204-E** — `decl_prior`'s `%hi/%lo` ARM HAS NEVER FIRED: the card's promised GLOBAL-TYPE row is 0 of 1,210 L21889 - **§204-CONFIRMED** — 30 reports that the index already answered L21941 - **§204-REJECTED** — sixteen, twice the previous record L22035 - **§205** — THE CHAINED ASSIGNMENT IS ITS OWN SCHEDULING DIAL: `*b = *a = v;` moves an argument copy that no local, no pin and no statement reorder will move (P31 S56) L22108 - **§206** — THE JTBL-CARVE DRAFTING IDIOM: the bounds check is the entry count, and an EMPTY case owns a slot (P31 S56) L22163 - **§207** — THE WAVE ab–ag HARVEST (P31 S58): 278 byte-banked notes → 25 laws, 103 self-reported no-gap L22232 - **§208** — TWO NAMED LOCALS FOR ONE RELOADED EXPRESSION BUY TWO ALLOCNOS — the naming granularity owns the REGISTER SPLIT, not just the load count (P31 S58) L22253 - **§209** — THE NARROW LOCAL IS A DIAL IN TWO OPPOSITE DIRECTIONS, AND §194-B's "≥2 `sh` STORES" BOUND IS BYTE-WRONG (P31 S58) L22302 - **§210** — THE SINGLE-BIT MASK IN A BOOLEAN TAIL: `andi K ; sltu $zero,v` vs `srl n ; andi 1` is a STATEMENT-SHAPE dial, not an operator choice (P31 S58) L22379 - **§211** — HOIST THE LOOP INIT ABOVE THE DOMINATING GUARD: it fills the guard's delay slot AND flips the counter/pointer register pair (P31 S58) L22425 - **§212** — THE WALKING CURSOR IS COUNTABLE: `*wp++` emits one `addiu` PER STORE, `wp[0..2]` emits one (P31 S58) L22484 - **§213** — INDEPENDENT SAME-BASE STORES: THE EMISSION ORDER IS A PERMUTATION OF SOURCE ORDER, AND THE PERMUTATION IS NOT ALWAYS THE IDENTITY (P31 S58) L22529 - **§214** — THE BANKED TWIN MAY BE A MACRO, A DELETED `.s`, OR A SEMANTIC INVERSE — six ways a ≥0.9 similarity lies (P31 S58) L22565 - **§215** — PIN ECONOMY: the twin's pins are NOT part of the shape, and §17's "pin every call-crossing value" is over-broad (P31 S58) L22611 - **§216** — DISTINCT ADJACENT SCALARS vs ONE ARRAY: one `lui` per access is the tell, and the array decl is UNUSABLE (P31 S58) L22667 - **§217** — DECODING A CALL'S STACK ARGUMENT SLOTS: `sw` at `0x10`/`0x14`/`0x18` are params 5/6/7 **(single observation — not yet cross-confirmed)** (P31 S58) L22707 - **§218** — A NARROW TYPE AT THE ABI BOUNDARY COSTS AN IN-PLACE `sll/sra` PAIR — on the RETURN as well as on the PARAMETER (P31 S58) L22733 - **§219** — COMPOUND `+=`, FULL ASSIGNMENT, AND AN EXPLICIT TEMP ARE THREE DIFFERENT SCHEDULES OF ONE READ-MODIFY-WRITE (P31 S58) L22762 - **§220** — THE PARAMETER ITSELF IS A REGALLOC DIAL: use it directly, prefer `s32` to `void*`, and place the save-copy AFTER the first call (P31 S58) L22789 - **§221** — A CONSTANT SHARED BY TWO STORES DIES AT THE CALL WHOSE DELAY SLOT REFILLS ITS REGISTER **(single observation — not yet cross-confirmed)** (P31 S58) L22829 - **§222** — SWITCH vs IF-CHAIN, PART 3: source arm order IS emission order, a leading EMPTY case buys the median split, and a 2-way dispatch with a shared post-block is a `switch` (P31 S58) L22845 - **§223** — READING A `jal` DELAY SLOT: the value in it was produced BEFORE the call, so it is NEVER that call's return (P31 S58) L22887 - **§224** — CROSS-JUMP: WRITE THE DUPLICATE, AND READ A SHARED DELAY SLOT AS THE MERGE SIGNATURE (P31 S58) L22949 - **§225** — THREE CONTROL-FLOW SHAPES NO STRUCTURED SPELLING REACHES (P31 S58) L22985 - **§226** — FRAME PADS: FOUR WAYS §162i1/§2429's DEAD-LOCAL LEVER MISFIRES (P31 S58) L23028 - **§227** — TYPE THE SOURCE BY THE **LOAD** WIDTH, NOT BY THE STORE WIDTH (P31 S58) L23069 - **§228** — READING THE DIVIDE, PART N: the off-by-one compare is `% K == 1`, and three more discriminators (P31 S58) L23089 - **§229** — THE ADDRESS IS A VALUE: NAMING IT MOVES THE `lui`/`addiu` PAIR — AND §L14410 SAYS THE OPPOSITE FOR A REASON (P31 S58) L23126 - **§230** — THE ANCHOR PROBE: with `%hi`/`%lo` masked, the surviving `addiu` deltas tell you which assignment was written first **(single observation — not yet cross-confirmed)** (P31 S58) L23175 - **§231** — TRANSCRIPTION AND SEMANTIC-READ HYGIENE: six ways the listing misleads (P31 S58) L23190 - **§232** — WHEN THE `jr` DELAY SLOT'S STORE STORES THE RETURN VALUE, TIE THEM WITH ONE PSEUDO **(single observation — not yet cross-confirmed)** (P31 S58) L23231 - **§30** — addendum (P31 S58) — the `/s` grant closes a SCHEDULING residual and a REGISTER residual with one edit L23267 - **§194-B** — addendum (P31 S58) — BOUND 2 is byte-wrong; see §209 Direction A L23287 - **§176-B** — addendum (P31 S58) — the misdiagnosis direction L23293 - **§165-40** — addendum (P31 S58) — the barrier goes at the COPY SITE, not inside the region it protects L23299 - **§164-63** — addendum (P31 S58) — the interposed asm works when the SECOND value is an ordinary assignment L23308 - **§193-A** — / §194-E addendum (P31 S58) — where the twin's body actually lives L23317 - **§233** — THE WAVE aa–bg HARVEST (P31 S58b): 1,101 byte-banked notes → 24 new laws, 21 addenda, ~700 already-covered L23323 - **§234** — CONSTANT MATERIALISATION: THE STORE LVALUE'S SIGNEDNESS PICKS `addiu` vs `li`/`ori` (P31 S58b) L23367 - **§235** — THE PHANTOM SYMBOL: A MASKED `MATCH` CAN CARRY A RELOCATION THAT DOES NOT EXIST (P31 S58b) L23408 - **§236** — THE DECLARATION LAYER IS THE DOMINANT BANK-BLOCKER: NINE WAYS A BYTE-PERFECT BODY FAILS THE GATE (P31 S58b) L23437 - **§237** — THE CAST-AT-CALL-SITE DECISION TABLE: WHAT §17a-1 FIXES, WHAT IT CANNOT, AND THE FOUR ESCAPES (P31 S58b) L23516 - **§238** — SAME NAME, DIFFERENT FUNCTION: THE OVERLAY-HOMONYM TRAP (P31 S58b) L23571 - **§239** — TWO-STATEMENT INTEGER-SPACE MATERIALISATION REORDERS `la` vs `sll`; AND THE PLUS-TREE OPERAND ORDER (P31 S58b) L23611 - **§240** — `A + K + B`: WRITE THE CONSTANT **BETWEEN** THE TWO RUNTIME TERMS (P31 S58b) L23645 - **§241** — THE FOLDED SIGN-EXTEND-AND-SCALE: `sll 16 ; sra (16 − log2 scale)` (P31 S58b) L23674 - **§242** — `*k` vs `<>n`: EXPRESSION SPELLING OWNS THE LOAD WIDTH AND THE ROUNDING CHAIN (P31 S58b) L23703 - **§243** — SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P31 S58b) L23731 - **§244** — `volatile` IS A COUNTING INSTRUMENT, A STORE-ORDER PIN, AND MUST SOMETIMES BE ASYMMETRIC (P31 S58b) L23761 - **§245** — THE CALL'S ARGUMENT LIST IS A SCHEDULING SLOT (P31 S58b) L23796 - **§246** — THREE-LIVE-VALUE SCAN LOOPS WANT ADDRESS-FROM-INDEX; AND TWO SYMBOLS CAN SHARE ONE giv (P31 S58b) L23843 - **§247** — TWO BRANCHES TO **ONE** LABEL MEANS THE SOURCE CONDITION IS NEGATED (P31 S58b) L23883 - **§248** — SPLIT THE LOAD FROM THE ARITHMETIC: A FUSED `g + K` DENIES THE CALLEE-SAVED REGISTER ITS DIRECT HOME (P31 S58b) L23917 - **§249** — THE SELF-ASSIGN, THE DEAD RE-ASSIGN, AND THE `+ zr` COPY: THREE WAYS TO MAKE A DELETED INSTRUCTION REAL (P31 S58b) L23941 - **§250** — `%hi/%lo` vs `lw`: THE EXTERN'S ARRAY-vs-SCALAR SHAPE DECIDES ADDRESS MATERIALISATION (P31 S58b) L23983 - **§251** — IMMEDIATE-SPELLING TRIGGERS: `+= 0xFF`, FULL-WIDTH `~K`, AND THE TWO-OR SPLIT (P31 S58b) L24020 - **§252** — THE GUARDED PRE-DECREMENT: `(x != 0) && (--x == 0)` (P31 S58b) L24042 - **§253** — POSTFIX `++` vs `+= 1` PICKS A DIFFERENT SCRATCH REGISTER **(single observation — not yet cross-confirmed)** (P31 S58b) L24063 - **§254** — THE DEAD PARAMETER IS A REGISTER-PLACEMENT TOOL **(single observation — not yet cross-confirmed)** (P31 S58b) L24077 - **§255** — THE EMPTY CASE, PART 2: FOUR TREE SHAPES IT BUYS (P31 S58b) L24089 - **§256** — GOTOS IN THE TARGET'S BLOCK ORDER REPRODUCE SWITCH PLACEMENT WITHOUT SWITCH'S SIDE EFFECTS (P31 S58b) L24124 - **§257** — THE DEAD-END LEDGER (P31 S58b): ELEVEN LEVERS THAT MEASURED NULL OR BACKFIRED L24161 - **§258** — ADDENDA TO EXISTING SECTIONS (P31 S58b) L24204 - **§30** — addendum (P31 S58b) — THE ANONYMOUS STRUCT MEMBER REF GRANTS `/s`, AND THAT IS A TWO-FOR-ONE L24209 - **§194-B** — / §209 addendum (P31 S58b) — TWO MORE INSTANCES, AND THE BOUND IS NOW REFUTED FOUR WAYS L24230 - **§202** — addendum (P31 S58b) — THE DEF-SIDE ALIAS ALSO CLEARS A RETURN+PARAM DOUBLE CONFLICT L24253 - **§205** — addendum (P31 S58b) — CHAINED ASSIGNMENT: N≥3 IS INNERMOST-FIRST, AND THE TEXT MIRRORS EMISSION L24264 - **§208** — addendum (P31 S58b) — IT SCALES TO SIX SITES, AND IT HAS AN EXACT INVERSE L24281 - **§210** — addendum (P31 S58b) — THREE CONFIRMED SPELLINGS OF THE BOOLEAN TAIL L24310 - **§211** — addendum (P31 S58b) — INIT PLACEMENT: FIVE MORE DIALS BEYOND THE GUARD HOIST L24326 - **§213** — addendum (P31 S58b) — THREE MORE PERMUTATION LAWS FOR INDEPENDENT SAME-BASE STORES L24369 - **§214** — addendum (P31 S58b) — FOUR MORE WAYS A HIGH-SIMILARITY TWIN LIES L24398 - **§215** — addendum (P31 S58b) — PIN ECONOMY, PART 2: NINE REFINEMENTS L24434 - **§217** — CROSS-CONFIRMED (P31 S58b) — AND THE INCOMING HOME SLOT IS THE MIRROR L24486 - **§220** — addendum (P31 S58b) — THE PARAMETER, NOT A COPY (SEVEN CARDS) L24503 - **§222** — addendum (P31 S58b) — IF-CHAIN vs SWITCH: THREE MORE DISCRIMINATORS L24528 - **§223** — addendum (P31 S58b) — FIVE MORE CONFIRMATIONS, AND THE CONSTANT-IN-`$v0` CASE L24546 - **§224** — addendum (P31 S58b) — CROSS-JUMP MERGES *CALLS*, AND THE DELAY SLOT IS THE DISCRIMINATOR L24570 - **§225** — addendum (P31 S58b) — THE GUARD-CLAUSE FINGERPRINT, AND THREE MORE SHAPES L24600 - **§226** — addendum (P31 S58b) — THE FRAME CATALOGUE: SEVEN MORE LEVERS, AND SLOT ORDER IS DECLARATION ORDER L24627 - **§229** — addendum (P31 S58b) — NAME IT **INSIDE** THE ARM L24672 - **§230** — CROSS-CONFIRMED (P31 S58b) L24683 - **§231** — addendum (P31 S58b) — FOUR MORE WAYS THE LISTING MISLEADS L24692 - **§232** — CROSS-CONFIRMED (P31 S58b) L24723 - **§259** — THE DISCARD LEDGER FOR THE aa–bg HARVEST (P31 S58b): WHAT WAS MINED AND REJECTED, AND WHY L24734 - **§260** — THE §154-A LEADING-ISLAND SPLIT: ONE CONFIG LINE, AND THE ISLAND PEELS FROM THE END (P31 S59, byte-proven) L24773 - **§260-A** — STAGE 2 IS PROVEN, AND THE WHOLE jtbl PIPELINE IS AUTOMATED AT THE GATE (P31 S59, same day) L24823 - **§261** — THE -O0 ORACLE: DERIVE THE OPT LEVEL FROM THE TARGET, AND `$fp` IS NOT THE TELL (P31 S59) L24859 - **§261a** — THE -O0 FRAME-RELOAD GRAMMAR: reload COUNT disambiguates the C spelling (P31 S59, byte-proven) L24885 - **§262** — A LANE'S YIELD IS ONLY A LANE FACT IF IT IS SIZE-MATCHED (P31 S59) L24907 - **§263** — A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCHEDULE (P31 S59, byte-proven) L24937 - **§264** — FOUR TELLS-LANE C RECIPES, EACH DRIVEN TO MATCH (P31 S59) L24978 - **§265** — THE VERBATIM-ASM BANK LANE: A FUNCTION NO -O2 C CAN EVER MATCH BANKS AS A RAW `__asm__` BODY (P31 S59b; two banked cards, two in-tree precedents) L25028 - **§266** — THE INERT-RIDER LAW: A LEVER IS ONLY CITABLE WHEN ITS SOLO REMOVAL BREAKS THE MATCH (P31 S59b; measured 4-of-8 on this batch) L25090 - **§267** — ADDENDA HARVESTED FROM WAVES at/bh/bk/bl (P31 S59b) L25130 - **ADD-1** — → §231 addendum (also cross-ref from §195-D) — THE MASKED-`jal` "MISSING CALL" ILLUSION L25138 - **ADD-2** — → §42a addendum — A SHARED CONSTANT *NAMED IN A LOCAL* ACROSS A `jal` IS AN ISO→TU DRIFT HAZARD; WRITE BARE LITERALS L25147 - **ADD-3** — → §236, item 10 — THE UN-DELETED `INCLUDE_ASM` STUB IS A DUPLICATE DEFINITION L25159 - **ADD-4** — → §225-3/-4 addendum — THE MIRROR ROW: VALUE-RETURN IN THE *TAKEN* ARM + TRAILING BARE `return 0` L25169 - **ADD-5** — → §1/I1 addendum — THE INVERTED RANGE TEST: `(u32)(x-lo) >= N` WITH THE ZERO-ARM AS THE TRAILING `else` L25183 - **ADD-6** — → §172b-1 / §264 addendum — SHIFT-AS-TEST: `(x << 16) != 0` TESTS THE LOW HALF WITHOUT TRUNCATING THE PSEUDO L25192 - **ADD-7** — → §256 addendum — THE SINGLE-GUARD GOTO: THEN-BLOCK OUT OF LINE AT THE TAIL L25201 - **ADD-8** — → §245 addendum — THE RE-TIE BARRIER PINS CALL-ARG SETUP TO SOURCE ORDER (construct 6) L25213 - **ADD-9** — → §213-3 / §217 addendum — ADJACENT PRE-CALL FRAME STORES ARE ONE AGGREGATE; THE UNESCAPED SCALAR NEIGHBOR IS DEAD-STORED L25225 - **ADD-10** — → §237 addendum (arity-evidence paragraph) — AN `la` PAIR ABOVE THE PROLOGUE `sw $ra` IS AN OUTGOING-ARGUMENT MATERIALISATION L25235 - **ADD-11** — → Cross-confirmation card block (per §259's standing instruction: confirmation, not news) L25244 - **§268** — A `register __asm__` PIN ON A CALL-CLOBBERED REGISTER IS HONORED EXACTLY WHEN THE PINNED RANGE CROSSES NO CALL (P31 S59c; three A/B'd cards, unifying §257-2's two) L25295 - **§269** — ADDENDA HARVESTED FROM WAVES ax/bm (P31 S59c) L25349 - **ADD-1** — → §257-8 addendum — THE INTERPOSED ASM'S `__volatile__` IS A PER-SHAPE DIAL, AND THE "NO-OUTPUT ASM IS IMPLICITLY VOLATILE" LORE IS BYTE-FALSE IN gcc-2.7.2 L25364 - **ADD-2** — → §236 addendum (item 1 corollary) — THE SILENT-SPLICE DECLARATION LADDER: WHERE THE TU HAS NO SPELLING, UNPROTOTYPED `()` BEATS THE FLEET VOTE L25377 - **ADD-3** — → §238 addendum — TWO BINARIES CAN EACH DEFINE THE SAME `func_` NAME WITH DIFFERENT BYTES, AND NO GENERATED REPORT SHOWS IT L25391 - **ADD-4** — → §167-37 addendum — THE FIFTH PRECONDITION, NOW MEASURED: DOWNSTREAM FIELD RE-READS ⇒ NAME IT **AND** STORE **INSIDE** THE ARM L25404 - **ADD-5** — → §267-ADD-6 / §172b-1 addendum — THE SIGN-TEST FACE: `(s16)v < 0` IS `sll 16` + `bgez` ON THE SHIFTED COPY, RAW STAYS LIVE — AND THE HALFWORD-ABS SHAPE NEEDS NO RITUAL L25421 - **ADD-6** — → §220-addendum — THE FOURTH FACE: WHEN THE NAMED COPY ITSELF BUYS THE EXTRA CALLEE-SAVED, PIN THE COPY TO `$16` L25437 - **ADD-7** — → §229 addendum — THE VALUE FACE: A LOOP-STORE CONSTANT SPELLED AS A LITERAL HAS NO SOURCE POSITION; NAME IT TO PIN THE PROLOGUE INIT ORDER L25453 - **ADD-8** — → §30a addendum — THE THIRD COLUMN: `*p++` IS ALSO SERIALIZED BUT BURNS `addiu`; ONLY CAST-PLUS GIVES SERIALIZATION *AND* DISPLACEMENT FOLDING L25466 - **ADD-9** — → §255 "AND CASE-BODY PLACEMENT" bound / §222-addendum-3 — ON A LARGE SPARSE TREE, BODIES FOLLOW **SOURCE** ORDER (measured by a one-word probe) L25481 - **ADD-10** — → Cross-confirmation card block (per §259's standing instruction: confirmation, not news) L25495 - **§3-1a.** — The §266 sweep — every solo-lever A/B run for this batch L25536 - **§270** — The A-prop 0-bank anatomy: a byte-correct body still needs FOUR layers to agree (P31 S59) L25569 - **§271** — Ordinal IMM pairing: text order is NOT emission order; emit CANDIDATES, let the oracle pick (P31 S59) L25587 - **§272** — The `(void)`-decl + empty-call wall: K&R the DEFINITION, not just the decls (P31 S59) L25609 - **§273** — A standalone compile is the WRONG oracle for a TU-destined draft (P31 S59) L25629 - **§274** — ADDENDA HARVESTED FROM 18 WAVES (P31 S60): 315 candidates, 255 already covered, 21 sharpenings, 3 new laws L25644 - **ADDENDUM** — to §179-C — the `.type NAME, @function` requirement L25650 - **ADDENDUM** — to §134 — a typedef defined BELOW the splice point is stripped anyway L25697 - **ADDENDUM** — to §179-D (GTE macro reference family) — `gte_SetRotMatrix`/`gte_SetTransMatrix` bodies L25727 - **ADDENDUM** — to §82 — struct copies must stay MEMBER-WISE, not block-moved, to match a spill-slot save L25753 - **ADDENDUM** — to §136d-1 (RC-12, the `$0`-add / opaque-copy family) — two symptoms beyond "compare reads the wrong register" L25791 - **ADDENDUM** — to §20 — a global declared as `T *` may itself BE the array base, not a pointer to dereference L25830 - **ADDENDUM** — to §1-I5 L25859 - **ADDENDUM** — to §164-51 L25924 - **ADDENDUM** — to §176-F5 L25940 - **ADDENDUM** — to §225 L25969 - **ADDENDUM** — to §224 — CROSS-JUMP: THE DUPLICATE CAN BE A PLAIN STORE, NOT ONLY A CALL L26014 - **ADDENDUM** — to §164-64 — AN EMPTY CLOBBER ON AN ARGUMENT REGISTER CAN BE THE DELIBERATE FIX, NOT JUST THE ACCIDENTAL BUG L26048 - **ADDENDUM** — to §244 — THE COMPILER BARRIER, NOT `volatile`, BLOCKS A STORE→LOAD HOIST THROUGH A POINTER-OFFSET FIELD L26079 - **ADDENDUM** — to §172b-1 — TWO PLACEMENT DIALS THE PROMOTION LAW DOESN'T NAME: NARROW THE COUNTER'S DECLARED TYPE, AND MOVE THE CAST INTO THE LOOP BODY L26110 - **ADDENDUM** — to §229 — A POINTER'S NUMBER OF USES DECIDES WHETHER ITS ADDRESS FOLDS OR SURVIVES A CALL L26157 - **ADDENDUM** — to §172b-4 — THE PLAIN CAST-DIVISION ALREADY PRODUCES THE PATTERN; DON'T HAND-ROLL THE BIAS, AND KEEP THE OPERAND WIDE L26202 - **ADDENDUM** — to §8 — WHEN `INCLUDE_RODATA` NEEDS A STANDALONE `.s` YOU CAN'T CREATE, CARRY THE FRAGMENT AS A FILE-SCOPE `__asm__` BLOB L26232 - **ADDENDUM** — to §226 — A DEAD LOCAL SCOPED TO A NESTED BLOCK BUYS A MID-FUNCTION `addiu sp` PAIR, NOT A FRAME-SIZE CHANGE L26275 - **ADDENDUM** — to §172a — RE-READING MEMORY (NOT NAMING A TEMP) IS WHAT KEEPS AN INCREMENT'S DELAY-SLOT FILL ALIVE L26314 - **ADDENDUM** — to §225 / §256 — A GOTO TO A SHARED SET-POINT PREVENTS IF-CONVERSION FROM COLLAPSING A LATER BRANCH TEST L26345 - **ADDENDUM** — to §211 — AN IN-LOOP ACCUMULATOR WANTS A CLOSED-FORM EXPRESSION WHEN THE TARGET REMATERIALISES ITS CONSTANT AFTER EVERY CALL L26387 - **§275** — THE LEFTOVER-REGISTER READ L26422 - **§276** — MIXED ADDRESS-EXPRESSION SPELLING FOR ADJACENT RELOCATABLE SYMBOLS IS A CSE-UNIFICATION DIAL, NOT JUST A BYTE-ENCODING CHOICE (P31 S60; `func_80180FE8`, ov_SC06_006, byte-proven) L26462 - **§277** — RETURN-TAIL C SPELLING PICKS THE DELAY-SLOT-FILL vs TRAILING-MOVE TOPOLOGY, AND A NARROWER SECOND VARIABLE KEEPS TWO PSEUDOS INSTEAD OF ONE (P31 S60; `func_801846F0` ov_SC03_104, `func_801A44C4` md_SC07_004, both byte-proven) L26571 - **§278** — ADDENDA HARVESTED FROM WAVE cf (P31 S60): 34 candidates, 13 already covered, 8 sharpenings, 4 new laws L26635 - **ADDENDUM** — to §74 (func_800D0E30, resident) L26643 - **ADDENDUM** — to §172b-4 (func_80185054, ov_SC03_097) L26679 - **ADDENDUM** — to §265 (func_8017DC80, ov_SC07_002) L26717 - **ADDENDUM** — to §17 (func_800CB794, md_MAIN_036) L26756 - **ADDENDUM** — to §162p (func_8017C120, ov_MAIN_012) L26804 - **ADDENDUM** — to §20 (~L1947) (func_80186AD0, ov_SC06_032) L26838 - **ADDENDUM** — to §164-56 (func_8017F644, ov_SC04_005) L26868 - **ADDENDUM** — to §237 (func_8017F7FC, ov_SC03_092) L26903 - **§279** — A `do/while (p < end)` LOOP UNDER AN ENTRY GUARD: the guard decides the compare, not the loop (P31 S60; wave cf, func_8017F2A4, ov_SC03_096, byte-proven) L26932 - **§NNN** — A DO-WHILE'S GUARD AND LATCH MUST REPEAT THE SAME BOUND EXPRESSION, NOT SHARE ONE LOCAL: THE REPETITION IS WHAT BUYS THE CSE'D COPY INTO A SECOND REGISTER (P31 S60; `func_8017F2A4`, ov_SC03_096, byte-proven 25/25) L26933 - **§280** — THE CURSOR'S DECLARED TYPE PICKS `sltu` vs `slt`: a `T*` bound test is UNSIGNED by C rule, an integer cursor is signed (P31 S60; wave cf, func_800CAE74, md_MAIN_031, byte-proven) L26966 - **§NNN** — A LOOP CURSOR'S C TYPE (POINTER vs PLAIN INTEGER) SELECTS `sltu` vs `slt` FOR ITS BOUND TEST, INDEPENDENT OF THE VALUES INVOLVED (P31 S60; `func_800CAE74`, md_MAIN_031, byte-proven; cross-confirmed same wave by `func_8017F2A4`, ov_SC03_096) L26967 - **§281** — GROUP COPY-THEN-RMW BY OPERATION KIND, NOT FIELD BY FIELD: sched1 does the interleaving, the source must not (P31 S60; wave cf, func_80180FB4, ov_SC03_111, byte-proven) L27000 - **§NNN** — A DEPENDENT COPY-THEN-RMW BLOCK MUST BE SOURCE-GROUPED BY OPERATION KIND, NOT BY FIELD, AND SCHED1 DOES THE INTERLEAVING (P31 S60; `func_80180FB4`, ov_SC03_111, byte-proven) L27001 - **§282** — gcc's OWN LOOP REVERSAL PUTS THE COUNTER INIT AFTER THE HOISTED MOVABLES — a position no hand-written down-count can reach (P31 S60; wave cf, func_80180DD8, ov_SC04_005, byte-proven) L27046 - **§NNN** — WRITE THE UP-COUNT LOOP: gcc's OWN REVERSAL PRODUCES A COUNTER-INIT INSTRUCTION THAT LANDS AFTER HOISTED MOVABLES, WHICH A HAND-WRITTEN DOWN-COUNT LOOP CANNOT REPRODUCE (P31 S60; `func_80180DD8`, ov_SC04_005, byte-proven) L27047 - **What** — I could not verify L27084 - **§283** — ADDENDA HARVESTED FROM THE 36-WAVE BATCH (P31 S60): 1,216 candidates, 859 already covered, 46 sharpenings, 9 new laws L27131 - **ADDENDUM** — to §164-75 — the fold-reassociation law also fires at a variable's INITIALIZER, not only a later expression L27141 - **ADDENDUM** — to §21 — the `bltz`+`slti` (or N-separate-compares) signed-range-split bullet is now CONFIRMED on three independent functions, and generalizes beyond `lbu`/u8 L27167 - **ADDENDUM** — to §202 — the DEF-SIDE ALIAS also resolves a function-vs-DATA-symbol identifier clash, not only a function-vs-function prototype clash L27213 - **ADDENDUM** — to §22 (`volatile`-qualified-global reload lever, cookbook ~L1922) — for a NON-constant, same-address double RMW, a plain memory clobber beats `volatile`, and `volatile` actively breaks a delay-slot fill L27236 - **ADDENDUM** — to §195-E — a goto-ladder's STORES must sit AT the labels, after the gotos, not inline before them L27255 - **ADDENDUM** — to §195-N — a GNU statement-expression slider must sit INSIDE the conditional arm's value position, not as a post-hoc barrier, to block the store-flag transform on a ternary chain L27317 - **ADDENDUM** — to §176-B2 — in a micro-function with no long/short lifetime asymmetry, BOTH contending pseudos need their own hard-register pin L27374 - **ADDENDUM** — to §199-G — A `default:` LABEL GROUPED ONTO THE LAST CASE REMOVES THE `j default` TAIL, EVEN THOUGH THE 2-NODE HEADER STAYS ALL-POSITIVE L27421 - **ADDENDUM** — to §220 — REFERENCING THE RAW PARAMETER (NO NAMED COPY, NOT EVEN A PIN) LETS THE CALLEE-SAVED SPILL LAND IN THE FIRST CALL'S OWN DELAY SLOT L27490 - **ADDENDUM** — to §252 — a `>=0`/`<0` split on an unconditionally-decremented value needs the POSTFIX operator INSIDE the branch condition, not a prior statement L27548 - **ADDENDUM** — to §215 — FIFTH SHAPE: reused mask constants across two call-free merge sites each get their own whole-function hard-register pin, and a shared sub-expression at the second site must be its own statement L27575 - **What** — I could not verify L27658 - **Harness-defect** — flags L27729 - **ADDENDUM** — to §167-40 (func_8017E2CC, ov_SC04_015 — wave dg) L27822 - **ADDENDUM** — to §20's cross-jump EXPLOIT bullet (func_8017E360, ov_SC05_007 — wave dg) L27842 - **ADDENDUM** — to §87 (func_801815F4 ov_SC06_032; corroborating func_801840DC ov_SC05_017, func_80189C68 ov_SC03_006 — wave dg) L27856 - **ADDENDUM** — to §153 / §236-5 (func_801A419C, md_SC07_003 — waves di and dl, corroborating; refutes a contradicted dj-wave card) L27875 - **ADDENDUM** — to §263 (func_801E83AC, md_SC04_029 — wave dj) L27895 - **ADDENDUM** — to §265 (func_8017D878, ov_SC03_107 — wave dj; corroborated by a REJECTED, contradicted card in wave dm — see closing) L27909 - **ADDENDUM** — to §6 (func_801811F0, ov_SC03_102 — waves dj and dl, corroborated by a self-reported "nothing new" dm-wave card) L27923 - **ADDENDUM** — to §195-G (func_80183BB0, ov_SC05_001 — wave dj) L27937 - **ADDENDUM** — to the zero-byte-asm-slider family (§47 / §148-C / §153) (func_800CB874, md_MAIN_040 — wave dj; open tension with a more cautious dm-wave card — see closing) L27965 - **ADDENDUM** — to §194-B (func_8017EB34, ov_SC03_117 — wave dk; distinct from the §74 co-pinning finding on the SAME function below) L27983 - **ADDENDUM** — to §74 (func_8017EB34, ov_SC03_117 — counter/clamp variant; wave dj, corroborated by dk/dl/dm cards on the same function) L28015 - **ADDENDUM** — §37 — merged into the §153/§236-5 entry above (func_801A419C, wave dl) L28035 - **ADDENDUM** — §6 — merged into the §6 entry above (func_801811F0, wave dl) L28043 - **ADDENDUM** — to §238 (func_80182CB4, ov_SC02_000 — wave dl) L28051 - **ADDENDUM** — to §137a (func_801684B4, ov_MAIN_012; corroborated independently by func_80189E68, func_8017FF9C, func_801822B4, func_8018DA8C — wave dl) L28069 - **ADDENDUM** — to §8c / §88d (func_8016AB6C, ov_MAIN_012 — wave dm) L28085 - **ADDENDUM** — to §73 / §30#2 (func_800D1984, resident — wave dm) L28099 - **What** — I could not verify L28115 - **ADDENDUM** — to §174 Law 4 (func_8017E9A8, ov_SC06_015) L28135 - **ADDENDUM** — the zero-emission-asm family gains a REF-SLIDER PLACEMENT LAW and a paired RESTORER (func_800CB900, md_MAIN_026) L28175 - **From** — ck + cl + cm L28227 - **ADDENDUM** — §NNN — sharpens §55a / §164-37 / §165-28 (switch-vs-tree cluster): A SHAPE THAT LOOKS LIKE A SWITCH DISPATCH CAN BE PLAIN NESTED `if`s WHOSE SHARED BODY WAS TRIPLICATED BY THE SOURCE AND THEN CROSS-JUMP-MERGED BACK DOWN L28233 - **ADDENDUM** — §NNN — sharpens §167-13's boundary: CHAINING TWO IDENTICAL SIDE-BY-SIDE STORES INTO ONE C ASSIGNMENT STATEMENT IS A MID-BLOCK SCHEDULING-PRIORITY DIAL, NOT ONLY A STORE-ORDER SPELLING L28270 - **ADDENDUM** — §NNN — sharpens §162a3/§60b: A `sll $v0,16 / sltiu $v0,1` ZERO-TEST OF A JUST-DECREMENTED HALFWORD IS A SIGNED-TEMP WIDTH TELL OUTSIDE ANY SWITCH/RANGE-TEST CONTEXT L28303 - **ADDENDUM** — §NNN — sharpens §37/§124 (unspecified-parameter-list family): DERIVE A CALLEE'S ARITY LOWER BOUND FROM ITS OWN `.s` STACK-ARGUMENT READS, NOT FROM ANY SINGLE CALL SITE L28336 - **ADDENDUM** — to §5a (func_80181F74, ov_SC03_112, wave cn) L28369 - **ADDENDUM** — to §265 (func_8017E26C, ov_SC04_016, wave cn) L28419 - **ADDENDUM** — to §42b (func_8018247C, ov_SC07_002, waves cu + cw) L28457 - **ADDENDUM** — to §199-F family (func_8017EFB0, ov_SC02_021, wave cu) L28488 - **ADDENDUM** — to §195-E (func_800CFC1C, md_MAIN_003, wave cv) L28543 - **ADDENDUM** — to §215 addendum (func_800CB2C8, md_MAIN_033, waves cv + cw) L28588 - **ADDENDUM** — to §236 item 4 (func_8017D268, ov_SC04_006, wave cw) L28644 - **What** — I could not verify L28673 - **Harness-defect** — flags (not idioms — flagged for the operator) L28705 - **ADDENDUM** — to §250 (func_8017D7CC, ov_SC03_115 — cx/cy/cz/dr) L28746 - **ADDENDUM** — to §225 (func_8017E190, ov_SC03_115 — cx/cy/cz) L28791 - **ADDENDUM** — to §45-A (func_8017F6A4, ov_SC02_016 — cy/cz) L28844 - **ADDENDUM** — to §249 (func_80182ED4, ov_SC04_004 — dp/dr/dt) L28888 - **ADDENDUM** — to §199-A (func_801816FC, ov_SC02_005 — dp/dr/dt) L28933 - **What** — I could not verify L28979 - **Harness-defect** — flags L29022 - **§284** — COMBINE CAN REASSOCIATE TWO SEQUENTIAL BITWISE-AND MASKS INTO ONE AGAINST THE PRE-MASK VALUE; AN ASM IN/OUT FENCE RIGHT AFTER THE FIRST MASK BLOCKS IT (P31, wave dd, `func_8018087C`, ov_SC04_020, byte-proven) (P31 S60; waves #, byte-proven) L29084 - **§285** — PRE-INITIALIZING A VARIABLE WITH A SHARED CONSTANT BEFORE A BRANCH MAKES BOTH ARMS OF THE FOLLOWING if/else DESTRUCTIVELY REUSE THE SAME DESTINATION REGISTER FOR THEIR BITWISE RESULT (byte-proven; `func_801811F0`, ov_SC03_102, independently rediscovered across waves di/dj) (P31 S60; waves #, byte-proven) L29088 - **§286** — FOLD A STATEMENT'S SIDE EFFECT INTO A COMMA-EXPRESSION IN AN ARGUMENT POSITION TO PLACE ITS RTL RELATIVE TO A CALL'S OWN DELAY SLOT (P31 S60/dj; `func_80180A88`, ov_SC06_010, byte-proven 411/411) (P31 S60; waves #, byte-proven) L29092 - **§287** — A STACK-FRAME HOLE BELOW TWO ADDRESS-TAKEN AGGREGATE LOCALS IS A LEADING PAD MEMBER OF ONE COMBINED STRUCT, NOT SEPARATE LOCALS OR A REGISTER PIN (P31 S60/dj; `func_8017D104`, ov_SC06_027, byte-proven 47/47) (P31 S60; waves #, byte-proven) L29096 - **§288** — A REGISTER PIN DECLARED UNINITIALIZED AND ASSIGNED ONLY AT ITS LATE, SOLE USE STILL FORCES THE FIXED REGISTER'S SAVE/RESTORE, AND THE ASSIGNMENT'S SOURCE POSITION CONTROLS WHERE THE VALUE MATERIALIZES (P31; `func_80186530`, ov_SC02_017, byte-proven, match_one MATCH re-verified) (P31 S60; waves #, byte-proven) L29100 - **§289** — an array local's address-taken base keeps every element's store alive, even though only one pointer escapes (`func_80189EFC`, ov_SC04_011) (P31 S60; waves #, byte-proven) L29104 - **§290** — A SINGLE STRENGTH-REDUCED GIV CAN DRIVE STORES TO SEVERAL DISTINCT RELOCATABLE SYMBOLS, EACH KEEPING ITS OWN `%hi`/`%lo` ANCHOR (P31 S60; waves #, byte-proven) L29108 - **§291** — THE DELAY-SLOT FALSE-VALUE: A CONDITIONAL BRANCH'S ZERO ARM MUST BE A FALL-THROUGH-ADJACENT BLOCK ENDING IN AN EXPLICIT JUMP, OR REORG CANNOT MATERIALIZE IT INSIDE THE BRANCH'S OWN DELAY SLOT (P31 S60; waves #, byte-proven) L29112 - **§292** — DECLARING A SYMBOL UPSTREAM OF AN ALREADY-BANKED SIBLING THAT RELIES ON THAT SYMBOL'S IMPLICIT (K&R) DECLARATION CAN SILENTLY REPROTOTYPE THE SIBLING'S OWN CALL SITE (P31 S60; waves #, byte-proven) L29116 - **§293** — THE "WHOLE-OBJECT GATE NEEDS EVERY SIBLING" LAW, WRITTEN PROPERLY: IT IS THREE LAWS, AND THE LOAD-BEARING ONE IS "THE BASELINE, NOT THE SIBLINGS" (P31 S61; byte-proven on 15 binaries in one night) L29119 - **§294** — ADDENDA HARVESTED FROM THE S61 GEN0/MAXTOK/MAIN/DEEPSEEK BATCH (waves ab8/ab16/ab24/ab32 · g0a–g0f · m0a · ds1/ds2): 99 candidates, 52 covered, 15 notes → 10 addenda, 29 notes → 5 new laws (§295–§299), 3 refuted L29169 - **ADDENDUM** — to §127 — the -O0 constant-offset fold keys on the MEMBER-ACCESS tree shape, not on the index being constant L29180 - **ADDENDUM** — to §261a — FOUR MORE -O0 DIALS BEYOND THE RELOAD COUNT (boot.c + md_MAIN_011/003, wave m0a/g0a/g0f, byte-proven) L29208 - **ADDENDUM** — to §264-3 — the explicit entry copy's BOUND: required exactly when the RAW value must outlive the call that consumes only the PROMOTED value L29258 - **ADDENDUM** — to §172b-1 (counter-type dial) — the sll's SOURCE REGISTER is the placement discriminator L29278 - **ADDENDUM** — to §276 — the SHARE direction: spell the second adjacent symbol RELATIVE to force ONE anchor L29290 - **ADDENDUM** — to §31's density-dummy dial (L2460/L2497) — the dose is TWO refs, and the dummy must sit where the loser is live-through L29306 - **ADDENDUM** — to §282 — the ADDRESS-GIV face: write the walked address INLINE so strength reduction births the induction register at loop.c's own insertion point L29323 - **ADDENDUM** — to §253 — SECOND byte-proven card (upgrade from single-observation), and the placement face: postfix-in-condition parks the RMW store in the branch delay slot L29346 - **ADDENDUM** — to the L1800 anchor-steer bullet — a DERIVED-POINTER local silently flips the merged giv's anchor END L29367 - **ADDENDUM** — to §31's asm→layout inference — TWO WIDTH-PAIR DISCRIMINATORS (same offset, different widths) L29390 - **§295** — THE KERNEL-TRAP STUB: §81's jr-DETECTOR WITHOUT A TABLE IS A PsyQ SYSCALL TRAMPOLINE — ROUTE TO §265, NEVER INTO THE CARVE CHAIN (P31 S61; wave m0a, 9 cards byte-proven; resolves §182's held cluster) L29411 - **§296** — THE FRAME CHECK OUTRANKS THE ATLAS LEVER: READ PROLOGUE/EPILOGUE BEFORE DRAFTING ANY C — A REAL TELL CAN LIVE INSIDE AN UNREACHABLE FRAGMENT (P31 S61; wave m0a, 16 cards byte-proven; extends §179-C) L29466 - **§297** — ONE GIV SERVING MIXED-WIDTH LOADS *AND* A STORE NEEDS STRUCT-MEMBER SPELLING; CAST/INDEX SPELLINGS RE-SPLIT THE PSEUDO (P31 S61; wave m0a, `func_800347C8`, main, byte-proven 31/31) L29509 - **§298** — THE SHARED-TAIL POSITION DIAGNOSTIC: A FOLDED TAIL *BETWEEN* SWITCH ARMS MEANS PER-ARM DUPLICATED STATEMENTS, NOT A HOISTED POST-SWITCH STATEMENT (P31 S61; waves g0e/g0f, `func_80181B68`, ov_SC06_016, byte-proven 68/68) L29552 - **§299** — TWO INDEPENDENT EXTRACTION CHAINS EMIT CONTIGUOUSLY INSIDE ONE EXPRESSION; ONLY A STATEMENT BOUNDARY MAKES THE SCHEDULER INTERLEAVE THEM (P31 S61; wave m0a, `func_8003A404`, main, byte-proven 8/8) L29577 - **§300** — S61 DISTILL BATCH NOTES (waves ab8/ab16/ab24/ab32 · g0a–g0f · m0a · ds1/ds2 · m0b) L29612 - **§301** — AN INTERNAL `j` CARRIES `R_MIPS_26 .text`: rtu/match_one "MATCH" COULD NOT SEE WHICH LABEL A `j` TAKES — FIXED (`jrel`), AND THE TWO DRAFT SHAPES IT HID (P31 S62 T1; byte-proven 2/2, negative-controlled over 3,297 stubs) L29681 - **§302** — A RED BINARY IS A DRIFTED SPEC, NOT A MYSTERY: THE THREE CARVE-STATE INVARIANTS AND HOW TO DERIVE EACH FROM THE BYTES (P31 S62 T2; five reds healed in one session, 5/5, +39 held banks) L29729 - **§303** — MODULE ISLAND TABLES: DERIVE THE PADS AT BUILD TIME, PEEL NOTHING — THE §154-A/§260 "island-pads"/"island-blocked" WALLS DISSOLVE (P31 S62 T3a; byte-proven md_SC03_076 func_801F0A9C + func_801F0F28, sha 9a165e36…) L29773 - **§304** — SELF-DEFINING RODATA: WHEN A FUNCTION'S `.s` IS THE ONLY OWNER OF THE DATA IT REFERENCES, THE C BODY MUST DEFINE IT (P31 S62 T3; byte-proven md_MAIN_011/func_800D04F4) L29809 - **§305** — "CARVE-REFUSED" AT GATE TIME IS THREE NAMED, DETERMINISTIC CLASSES — NONE OF THEM A CARVE (P31 S62 T3; 28 resolver drafts autopsied 28/28) L29832 - **§306** — A HAZARD `nop` IN FRONT OF A DIV-RESULT STORE IS A STATEMENT-ORDER DEFECT: THE INDEPENDENT TRAILING STATEMENT MUST BE WRITTEN *BEFORE* THE DIVISION-CONSUMING ONE (P31 S62 T4; byte-proven func_8017E7D0) L29884 - **§306a** — T4 DISTILL ADDENDA (P31 S62; four byte-proven refinements to existing laws, verified against the book by an independent verifier; each names its parent §) L29911 - **§307** — THE BRUTE-FORCE-THE-STATEMENT-ORDERS LEVER HAS A BOUND: A FAN-OUT COPY'S PRIORITY IS SCHEDULER-INTERNAL, AND NO SOURCE ORDER REACHES IT (P31 S63; byte-evidenced NEGATIVE result, main wave) L29949 - **§308** — A NAIVE FINAL `goto` DELETES THE TARGET'S `j`+`nop`: SPELL THAT EDGE AS A TAUTOLOGICAL `if` SO cse FOLDS IT TO AN UNCONDITIONAL JUMP *AFTER* jump1's WINDOW HAS CLOSED (P31 S63 t5e-t5i; byte-proven func_80180808) L29993 - **§308a** — A TWO-INSTRUCTION ARM STUB LAID *BEFORE* THE OTHER ARM AND REACHED ONLY BY A TRAILING `j` SURVIVES ONLY BEHIND A SHORT-CIRCUIT `&&` GUARD PLUS A SEPARATE RE-TEST; EVERY PLAIN if/else-if, GOTO-LADDER, NESTED-INVERTED-if AND 1-/2-NODE switch LETS jump1 INLINE IT (−2 ins) (P31 S63 t5e-t5i; byte-proven func_80180808) L30022 - **§309** — A FRAME-ADDRESS EQUIVALENCE RIDES *THROUGH* A SKIPPABLE CONDITIONAL: POST-`if` STORES FOLD ONTO `$sp` UNLESS THE `if` SURVIVES jump1 AS A BARRIER-PRECEDED DIAMOND (P31 S63 t5e-t5i; byte-proven func_801812AC) L30056 - **§310** — A TWO-OPERAND `subu`'s DESTINATION TIES ONLY TO THE OPERAND THAT IS BLOCK-LOCAL: SHARE *ONE* SIDE ACROSS BLOCKS TO CHOOSE WHICH OPERAND'S REGISTER IT LANDS IN (P31 S64 t5j-t5m; byte-proven func_8017F578) L30096 - **§311** — A JOIN-BLOCK STORE CAN SWALLOW AN ARM'S LOAD-DELAY `nop`: WHEN ONE ARM'S VALUE COMES FROM A LOAD, DUPLICATE THE STORE INTO BOTH ARMS (P31 S64 t5j-t5m; byte-proven func_8017D7E0) L30127 - **§312** — A BARE RELATIONAL IN AN `if` GIVES THE `slt` NO TARGET: ASSIGN THE COMPARISON INTO THE ALREADY-ALLOCATED VARIABLE TO CHOOSE ITS DESTINATION REGISTER, AT ZERO INSTRUCTION COST (P31 S64 t5o-t5r; byte-proven func_80180DCC) L30169 - **§313** — A RECOVERY STAGE THAT REWRITES THE DRAFT IS PART OF THE MEASUREMENT: GATE THE RAW DRAFT FIRST, THE REWRITE ONLY AS A FALLBACK (P31 S65 t5r-recovery; byte-proven func_8017BEBC, 246 ins) L30195 - **§314** — AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES ALL COMPILE BYTE-IDENTICAL, AND NONE REACHES A TARGET THAT SINKS THE NEGATIVE ARM INTO AN OUT-OF-LINE BLOCK (P31 S65; UNPROVEN, gate-refused draft func_80181720) L30251 - **§315** — ALL-CONSTANT AGGREGATE FILL: THE EMISSION ORDER IS SHARED-LITERAL GROUPS × DESCENDING INDEX, WITH PAIRED SUB-FIELDS INTERLEAVED (P31 S66; ⚠ **UNPROVEN** — `func_80185214`, gate-REFUSED draft, closeness 32→5, never MATCHed) L30283 - **§316** — A GUARD CHAIN THAT GATES A SUCCESS BLOCK ROUTES ITS FAILING GUARDS TO A LABEL **BEFORE** THE FLAG TEST, NOT TO THE EPILOGUE: THE RESIDUAL IS IMM-OFFSET-ONLY AT AN EXACTLY EQUAL INSTRUCTION COUNT (P31 S66; byte-proven func_80180284) L30320 - **§317** — A NARROW STRUCT-FIELD STORE TURNS A *LIVE SImode VARIABLE* INTO A `li`: convert.c's TRUNCATION-DISTRIBUTION MINTS A HImode COPY, AND cse REWRITES THAT COPY'S SOURCE TO THE EQUIVALENT CONSTANT (P31 S66; byte-proven func_8017EF94) L30354 - **§318** — A UNARY MINUS STORED BACK INTO THE SAME HALFWORD IS COMPUTED IN HImode, SO ITS LOAD IS `lhu`; AN INLINE `(s32)` CAST IS PROVABLY INERT AND ONLY A NAMED SImode LOCAL BUYS `lh` (P31 S66; `func_8018568C`, ov_SC03_028 — **`match_one`-only, NOT byte-gated**) L30388 - **Addendum** — §148-A2 addendum — the empty-asm insn_count dial (N bare `__asm__("")` = exactly +N real i (func_801835B0) L30463 - **Addendum** — §312 addendum — naming the compare's dest is the vehicle, the hard-reg pin on that name is (func_80184A68) L30471 - **Addendum** — subu/sh dest reuses a pinned operand's dying register (func_8017F498) L30475 - **Addendum** — Addendum to §261a — the -O0 increment-operator copy also fires on the POINTER LOCAL itself (func_8001212C) L30483 - **Addendum** — Addendum to §164-XX/§165-06 — at -O0 a bare local IS a memory lvalue, so the increment ope (func_8001212C) L30487 - **Addendum** — Addendum to §312 — the compare's named destination must itself carry a hard register: nami (func_80184A68) L30491 - **Addendum** — Addendum to §167-12 — the single-operand volatile keepalive, re-confirmed at a gated MATCH (func_8017F498) L30495 - **Addendum** — A §5a BARRIER SITS IN REORG'S FILL WINDOW TOO, SO ITS SIDE OF THE INNER `goto` IS CHOSEN P (func_80181D08) L30499 - **Addendum** — Loop sentinel wrongly hoisted out of a call-containing loop (func_801835B0) L30503 - **Addendum** — Reversed-operand sltu then xori 1 means unsigned less-or-equal (func_8017D948) L30515 - **Addendum** — Chained *2*2 array index folds to one copy;sll, not two (func_80011380) L30519 - **Addendum** — REGALLOC-PERM: one memory-clobber barrier insufficient — stack two (func_80182D1C) L30533 - **Addendum** — Last if-arm's nested if un-fuses the whole preceding chain (func_8017FE88) L30539 - **Addendum** — A plain global-RMW triplet is a fillable scheduling window too — zero drift, and the fille (func_800CB058) L30543 - **Addendum** — Shared arithmetic that CONSUMES the selected value cannot be hoisted above the `if` — narr (func_80183DA0) L30547 - **REFUTED** — claim — Dead call-arg register address-fold: no C-level lever found (func_8017ECAC) L30551 - **Addendum** — Truncating assign must be the lazy `||` operand, not hoisted (func_80012B58) L30570 - **Addendum** — WIDTH sh!=lhu residual: hoist a sibling field's load first (func_80182538) L30582 - **REFUTED** — claim — REFUTED — the `j`-slot store is an ASPSX macro-expansion hop, not a cse split; maspsx hard (func_8005DBD8) L30599 - **§NNN** — REFUTED: A `sw %lo(SYM)($at)` IN A JUMP'S DELAY SLOT WITH ITS `lui $at` STRANDED ABOVE THE JUMP IS **ASSEMBLER** MACRO-EXPANSION, NOT A cse SPLIT — OUR PIPELINE CANNOT EMIT IT FROM ANY C, AND THE +1 IS A `maspsx` GAP (P31 S66 round2, `func_8005DBD8`, byte-probed) L30601 - **Addendum** — Static local_type blocker survives typedef removal — it's textual (func_801588CC) L30625 - **Addendum** — Orphan sh triplet to $sp is a write-only local array (func_8017F274) L30632 - **Addendum** — REGALLOC-PERM: the store reads the narrow copy's register — split the one narrow local by (func_801838CC) L30636 - **Addendum** — REGALLOC-PERM store in a cross-jump shared tail defeats §137's barrier (func_8017EDE8) L30640 - **Addendum** — LENGTH-DRIFT nop clears when offset math precedes the symbol launder (func_80039B20) L30644 - **REFUTED** — claim — REFUTED: a block shared across TWO switch statements is ordinary forward cross_jump, not a (func_8001B0D4) L30648 - **§NNN** — REFUTED: A CODE BLOCK SHARED BY TWO *DIFFERENT* SWITCH STATEMENTS IS ORDINARY FORWARD `cross_jump`, NOT A SHAPE C CANNOT REACH — §164-11 ALREADY BANKS IT FROM PLAIN NESTED SWITCHES (P31 S66 round3, `func_8001B0D4`; ⚠ **UNPROVEN** — gate-REFUSED draft, closeness 68→45, never MATCHed) L30650 - **RETRIEVAL** — FAILURES this round (not new knowledge — a RETRIEVAL defect) L30666 - **§319** — N TEXTUALLY DUPLICATED `return v;` TAILS PUT THE LAST `or` AND THE RETURN-REGISTER MOVE IN ONE BASIC BLOCK, SO combine FOLDS THEM AND YOU ARE EXACTLY −1; A `goto done;` JOIN WITH REAL INCOMING EDGES RESTORES THE SEPARATE `addu $v0,$a0,$zero` (P31 S66 round4; byte-proven func_801809F0) L30681 - **Addendum** — Masked-OR field-merge plateaus at close=10; bitfield store clears it (func_8017FD64) L30710 - **Addendum** — Integer-space address arithmetic is a THIRD no-movable spelling, and a distant `SYM[0]` re (func_8017D89C) L30714 - **Addendum** — LENGTH-DRIFT −1 on a coalesced-away copy: a CALLER-SAVED SOURCE pin can resurrect it, boun (func_8017D72C) L30724 - **Addendum** — Return-0 statement order: extra j+move vs delay-slot fusion (func_8018BB50) L30731 - **Addendum** — Counter zero in the DECLARATION slot, empty for-init — the prologue SHIFT-DRIFT/+K face of (func_8018275C) L30735 - **Addendum** — The dying-pinned-register reuse on a sign-extend chain: route every later read through a s (func_80186868) L30753 - **Addendum** — Volatile zero-byte asm slider scrambles the prologue schedule (func_801814B0) L30757 - **Addendum** — Register-pinned helper pointer local regresses closeness; use plain local (func_80013CFC) L30767 - **Addendum** — §194-A addendum — the bare/colon-less fence measured NULL and only the "memory"-clobber fo (func_8017E464) L30771 - **RETRIEVAL** — FAILURES this round L30775 - **§320** — THE §43 "RETURN-TYPE FLIP PAIR" IS **NOT** TU-EDIT-REQUIRED: THREE DRAFT-ONLY ESCAPES (P31 S66; byte-proven func_800CCBC0, func_800D30D0, func_800D2A24) L30796 - **§321** — FILE-SCOPE DUPLICATE ANONYMOUS-STRUCT TYPEDEFS ARE A HARD ERROR; THE SAME TEXT AT BLOCK SCOPE IS A WARNING (P31 S66; byte-proven func_80180728, func_8017F9F8, func_8017E07C) L30855 - **§322** — A PROBE THAT ANSWERS A *NECESSARY BUT NOT SUFFICIENT* QUESTION WILL PRICE BLOCKED WORK AS FREE: RUN THE REAL PLANNER WHEN THE PLANNER IS PURE (P31 S67; measured, 96 of 159 open jtbl functions) L30876 - **§323** — CARRYING FILE-SCOPE TYPES BETWEEN SPLIT TUs: DEDUPE BY NAME, SKIP HEADER-PROVIDED, AND THE GUARD TRAVELS WITH THE BLOCK (P31 S67; ov_SC02_000, two defects fixed, one open) L30906 - **§323a** — R53, TWICE IN ONE HOUR: A FAILED BUILD LEAVES THE PREVIOUS BINARY, AND `sha1sum` READS GREEN (P31 S67) L30936 - **§323b** — A SCRIPT THAT PARSES argv AT IMPORT CANNOT BE SHARED; EXTRACT THE PREDICATE, DO NOT COPY IT (P31 S67) L30945 - **§324** — THE ARITY-SELF WALL NEEDS *BOTH* HALVES: A K&R DEFINITION **AND** A NO-PROTO TU DECL; EITHER ALONE STILL FAILS (P31 S67; rtu-proven ov_SC01_005/func_8017FBCC, 14 instances measured) L30959 - **§325** — REGALLOC: A SHARED SMALL CONSTANT STORED TWICE IN THE PRE-LOOP BLOCK IS A *local-alloc* $s-OCCUPANT THAT STEALS THE ARGUMENT ALLOCNO'S REGISTER — PIN THE ARGUMENT-DERIVED LOCAL, NOT THE CONSTANT (P31 S67; byte-proven ov_SC03_119/func_80184F18, 153 ins) L30997 - **§326** — DEFEATING ADDRESS-CSE: SPELL TWO READS OF THE SAME HALFWORD DIFFERENTLY AND GCC CANNOT SHARE THE ADDRESS (P31 S67; byte-proven ov_SC01_077/func_8017FAAC, 154 ins) L31010 - **§327** — A RANGE TEST MUST BE HImode: WITH `s32` + A `(u16)` CAST GCC *PROVES* THE MASK REDUNDANT AND DROPS THE `andi` (P31 S67; byte-proven ov_SC02_039/func_8017EC34, 154 ins) L31019 - **§328** — THE VOLATILE ALIAS MUST BE AN *OBJECT*, NOT A CAST: `*(volatile s32*)&sym` UNFOLDS %lo INTO A SEPARATE `addiu` (P31 S67; byte-proven ov_SC07_007/func_80181B8C, 156 ins, NEW LAW) L31034 - **§329** — fold-const NARROWS `(int)s16_var & 0xFFF` ONTO THE *RAW HImode PSEUDO*, BREAKING THE SIGN-EXTEND/MASK REGISTER TIE — A ZERO-BYTE WIDENING TEMP RESTORES IT (P31 S67; byte-proven ov_SC01_084/func_80183244, 157 ins) L31044 - **§330** — THE NEIGHBOUR-SHAPE LEVER IS THE CHEAPEST FIRST MOVE, AND IT IS UNDER-USED (P31 S67; four independent instances in one 20-function wave) L31052 - **§331** — OPEN GAP: NO LEVER ELIMINATES AN *UNWANTED DUPLICATE* COPY AT A BRANCH-TARGET BLOCK HEAD (P31 S67; main/func_80013154, closeness 12, NOT solved) L31069 - **§332** — THE maspsx `la`-IN-A-DELAY-SLOT GAP: gcc EMITS A SYMBOLIC ADDRESS LOAD AS ONE ATOMIC length-2 INSN, SO IT CAN NEVER FILL A JUMP DELAY SLOT — 6 FUNCTIONS FLEET-WIDE, NONE BANKABLE FROM C (P31 S67; byte-traced main/func_80062144, func_8005DBD8) L31081 - **§332a** — MAIN'S RESIDUAL FRONTIER IS CONTAMINATED WITH TOOLCHAIN WALLS: 4 OF 7 IN ONE WAVE (P31 S67, measured) L31106 - **§333** — FRAME SIZE IS SET BY *DECLARED* AGGREGATES, NOT USED ONES: AN UNREFERENCED TRAILING LOCAL IS A REAL DIAL (P31 S67; three independent byte-proven instances in one wave) L31125 - **§334** — A RELOAD SPILL SLOT IS ROUNDED TO `BIGGEST_ALIGNMENT` (8B), SO ONE SPILLED 4-BYTE PSEUDO CAN GROW THE FRAME BY 16 (P31 S67; byte-proven ov_SC05_008/func_8017DF68, 82 -> 53 residual) L31139 - **§335** — AN `extern u16 A[]` READ AT A VARIABLE SUBSCRIPT ALLOCATES DEAD STACK TEMPS (~8B PER ACCESS) THAT INFLATE THE FRAME WHILE EMITTING ZERO EXTRA INSTRUCTIONS (P31 S67; byte-proven ov_SC06_025/func_8017EA74, closeness 141 -> 20) L31150 - **§336** — THE §5a CROSS-JUMP BARRIER GOES AT THE *BOTTOM* OF THE TWIN, NOT THE TOP: `find_cross_jump` WALKS BACKWARD FROM THE CONVERGING JUMP (P31 S67; byte-proven ov_SC05_001/func_80183C9C) L31159 - **§337** — THE CC1-ONLY BLOCKER CLASS: `blocker_probe`'s STATIC ORACLE REPORTS "none" AND THE WHOLE-TU COMPILE STILL FAILS (P31 S67; ov_SC04_011/func_801827DC, md_MAIN_027/func_800CB4A4) L31167 - **§338** — `jtbl_carve._sltiu_bounds` MISREADS A NON-SWITCH `sltiu` AS A BOUNDS CHECK, OVER-SPANNING THE TABLE (P31 S67; ov_SC06_022/func_80185B80, byte-diagnosed) L31181 - **§339** — A 2-CASE SWITCH OMITS THE LOW-BOUND RANGE TEST, SO THE PRESENCE OF `slti/bnez` BETWEEN THE `beq`s IS A **COUNT TELL** FOR A THIRD CASE NODE (P31 S67; byte-proven ov_SC02_005/func_80190538, 197 ins) L31191 - **§340** — §194-K COROLLARY: **FLIP THE FALSE EDGE YOU CANNOT DELETE.** A "scheduler" residual can be sched.c's ALIAS ORACLE emitting a FALSE true-dependence; source order chooses its DIRECTION (P31 S67; byte-proven ov_SC03_107/func_8017CF48, 10 -> 0 in one compile, zero bytes) L31219 - **§341** — AN HImode STORE TEMP REWEIGHTS A sched2 TIE-BREAK THAT NO STATEMENT ORDER CAN REACH (P31 S67; byte-proven ov_SC03_006/func_801823B8, last 4 ins) L31253 - **§342** — A `void *` PARAMETER CAST TO ITS REAL TYPE IN A LOCAL IS **NOT** BYTE-NEUTRAL WHEN A LATER PARAMETER ALSO NEEDS A CALLEE-SAVED REGISTER (P31 S67; byte-proven ov_SC07_002/func_80181394, NEW LAW) L31267 - **§343** — `decl_prior`'s FLEET MAJORITY CAN BE WRONG ABOUT THE TRUE SIGNATURE — READ THE RIVALS, NOT JUST THE WINNER (P31 S67; measured on func_8012BD14 / func_8012D624 / func_80143C74) L31284 - **§344** — RAISE A BIV'S global_alloc PRIORITY WITH A ZERO-BYTE REFERENCE INSTEAD OF PINNING IT; PINNING THE COUNTER KILLS LSR ENTIRELY (P31 S67; byte-proven ov_SC03_121/func_80180E64, 222 ins) L31306 - **§345** — A VOLATILE **STORE** EVICTS THE MEM FROM cse AND KEEPS `sh`; A VOLATILE **LOAD** BLOCKS combine AND DEGRADES `lh` INTO `lhu+sll+sra` (P31 S67; byte-proven ov_SC01_084/func_80181A7C) L31322 - **§346** — `c ? X : -X` TAKES expand_expr's COND_EXPR **SINGLETON** PATH (copy, then negate IN PLACE) — AN if/else STATEMENT GIVES THE TWO-ARM FORM (P31 S67; byte-proven ov_SC03_102/func_80180C38, closed the last instruction) L31332 - **§347** — LOOP REGISTER ASSIGNMENT IS A **DECLARATION-ORDER + LIVE-RANGE** DIAL: FIVE COMPOSABLE LEVERS, 178 -> 0 (P31 S67; byte-proven ov_SC06_029/func_8017EF34, 243 ins) L31352 - **§347-addendum** — A THIRD INSTANCE, AND THE SHARPEST STATEMENT OF THE RULE (md_MAIN_025/func_800CB300, 243 ins) L31382 - **§343-addendum** — SECOND INSTANCE OF THE WRONG-MAJORITY DECL (same function) L31400 - **§348** — THE BASE SPELLING PICKS THE ADDRESSING MODE: A SYMBOL GIVES THE 3-INSN `lui/%lo` FORM, A POINTER VARIABLE GIVES THE 2-INSN `addu/lw` FORM (P31 S67; byte-proven ov_SC07_007/func_80182184, 264 -> 30 on this row alone) L31407 - **§349** — RE-ASSIGN A BASE POINTER AT THE END OF THE LOOP BODY TO MAKE `n_times_set > 1` — THAT KILLS BOTH loop.c's INVARIANT HOIST **AND** THE ADDRESS GIV, REPRODUCING A REMATERIALISED `addiu $aN,$sp,K` (P31 S67; byte-proven ov_SC06_029/func_801804C8, 255 ins) L31424 - **§350** — A ZERO-BYTE RE-TIE SETS `reg_n_sets=2`, WHICH KILLS sched1's `birthing_insn_p` LAUNCH_PRIORITY BOOST — THE MECHANISM BEHIND "MY ADDS ARE GLUED TO THEIR STORES" (P31 S67; byte-proven ov_SC04_011/func_80180B24, 215 ins) L31445 - **§351** — `/s` (MEM_IN_STRUCT_P) IS A DIAL YOU CHOOSE PER ACCESS: A COMPONENT_REF GRANTS IT AND LETS cse KEEP AN INDEX ACROSS THE STORE; A PLAIN CAST DENIES IT (P31 S67; ov_SC01_000/func_8017DD04, 186 -> 5) L31466 - **§352** — ⚠ TWO IDENTICAL `__asm__ __volatile__("")` BARRIERS **MERGE WITH EACH OTHER** — SPELL THE SECOND ONE DIFFERENTLY (P31 S67; byte-proven resident/func_800D128C, measured closeness 105 when they merged) L31496 - **§353** — USE `-fno-thread-jumps` AS AN **ORACLE** TO PROVE A RESIDUAL IS thread_jumps, THEN LAUNDER THE *VALUE* TO KEEP A DEAD RE-TEST (P31 S67; byte-proven ov_SC01_008/func_8017EC68, 279 ins) L31529 - **§354** — THE giv **WORTH-WHILE TEST** IS A DIAL: RE-ASSOCIATE THE ADDEND INTO THE INDEX TERM AND `strength_reduce` DECLINES (P31 S68; byte-proven ov_SC03_105/func_801824CC, 320 ins, 201 → 5 → 0) L31552 - **§355** — A REMAPPED SIBLING'S **SOURCE BIAS IS NOT ITS EMITTED BIAS** — DO NOT HAND-SHIFT OFFSETS TO MATCH THE ASM (P31 S68; byte-proven ov_SC07_007/func_8013DD68, 187/187 in 2 iterations) L31570 - **§356** — MEASURE A DRAFT IN THE TU IT WILL LIVE IN, NOT IN THE STANDALONE PROBE (P31 S68; measured over 47 stored drafts) L31584 - **§357** — ONE STRUCT POINTER, NOT TWO: A SECOND SOURCE VARIABLE BUILDS A THIRD IV (P31 S68; byte-proven ov_SC06_029/func_80181DF8, 335 ins, 330 → 13) L31599 - **§358** — (sharpens §333) — AN **UNREFERENCED** FIXED-SIZE AGGREGATE LOCAL IS LOAD-BEARING (P31 S68; same function) L31608 - **§359** — (§43-adjacent) — SPELL A SIGN-WIDEN AS AN EXPLICIT TWO-STEP, FUNCTION-SCOPED TEMP (P31 S68; byte-proven main/func_80012B58, 69 ins, 58 → 3 → 0) L31618 - **§360** — THE "COMPILER FOUND A SHORTER EQUIVALENT THAN THE TARGET" PAIR (P31 S68; main/func_800241C0, 68 ins, 68 → 19) L31631 - **§361** — ★ — A LOOP-TAIL BYTE SIGNATURE THAT NAMES ITS SOURCE SHAPE — AND THE LAW THAT A "SCHEDULING TIE" MAY BE YOUR OWN EARLIER LEVER (P31 S68; byte-proven main/func_800241C0, fable escalation, 19 → 0 in 3 iterations) L31647 - **§362** — TWO TRAPS WHEN A CARVE MOVES A STUB INTO THE `-O0` TU (P31 S68; byte-proven, 6 fns / 2,547 ins across ov_MAIN_012 / ov_SC02_037 / ov_SC03_107) L31692 - **§363** — ★★ — THE OVERLAY-LAYOUT ASSUMPTION IS A SYSTEMIC BUG CLASS, AND `main` IS THE EXCEPTION THAT FINDS IT (P31 S68; six instances, four of them in one session) L31715 - **§364** — ★ — THE libgpu `P_TAG` BITFIELD SPELLING IS **OPT-LEVEL DEPENDENT** (P31 S68; two functions, opposite verdicts, same session) L31744 - **§365** — PIN **BOTH** MASKS OR NEITHER (P31 S68; ov_SC01_000/func_8017E594, 357 ins) L31758 - **§366** — ★★ — `group_case_nodes` MERGES **STACKED CONSECUTIVE** CASE LABELS: GIVE EVERY CASE ITS OWN BODY (P31 S68; ov_SC01_001/func_8017EC28, **first-try MATCH 360/360**, 96/96 relocs audited) L31767 - **§367** — RECONCILING A DECL CONFLICT BETWEEN TWO DRAFTS FOR THE SAME TU (P31 S68; main) L31787 - **§368** — ★★★ — THE **RELOAD-REMAT CONSTANT**: REACH A REGISTER NO PIN CAN REACH (P31 S68; ov_SC03_105/func_80187A30, 339 ins, fable escalation closed 8 → 0 in ONE edit) L31802 - **§369** — REUSE THE **COMPARE CONSTANT'S OWN VARIABLE** FOR A MASK THAT KEEPS COALESCING (P31 S68; md_SC07_004/func_801AEC38, 365 ins) L31832 - **§370** — ★★ — A **HARD BOUND** FROM sched.c, AND THE reorg SLOT-STEAL DIAGNOSTIC (P31 S68; main/func_8001BC6C, 69 ins, NOT closed — 33 → 28 over ~45 measured compiles) L31848 - **§371** — ★★ — CARVING A **SINGLE-OBJECT MODULE BINARY**, AND THE spimdisasm RODATA-MIGRATION TRAP THAT FOLLOWS (P31 S68; byte-proven md_MAIN_003, func_800D0D6C 345 ins) L31900 - **§372** — ★★★ — THE **COPY-CAPTURE PAIR**, AND THE ONE ZERO-BYTE EDIT THAT DEFEATS BOTH (P31 S68; byte-proven main/func_8003491C, 78 ins, fable escalation closed 5 → 0) L31949 - **§373** — ★★★ — THE **DEAD-RESET CSE-BREAKER**, THE PIN THAT BREAKS A sched2 ANTI-DEP, AND WHY AN `asm` CAN NEVER RAISE PRIORITY (P31 S68; byte-proven ov_SC06_010/func_8017E764, 438 ins, fable escalation closed 8 → 0) L31987 - **§3-1.** — DEAD-RESET CSE-BREAKER — the zero-footprint replacement for a §195-I asm re-tie L31989 - **§3-2.** — A REGISTER PIN THAT DELETES A sched2 ANTI-DEPENDENCE L32007 - **§3-3.** — HARD FACT FOR THE SCHEDULING MAP — an `asm` ALWAYS has priority 1 L32020 - **CROSS-REFERENCE** — TO §370 — checked and found INAPPLICABLE here, which is the point L32028 - **§374** — A `register … __asm__("$30")` RESERVATION IS **NOT HONOURED** BY `move_movables` (P31 S68; main/func_80015608, 86 ins) L32036 - **§375** — AN `$a0`-`$a3` PIN USED LATE RELOCATES AN **EARLIER** OUTGOING-CALL USE OF THAT REGISTER (P31 S68; main/func_8005F0C8, 88 ins) L32052 - **§376** — ★★★ — A STANDALONE `match_one` CLOSENESS OF 0 IS A CLAIM ABOUT THE **BODY**, NEVER ABOUT THE **TU** (P31 S69; measured 0/28) L32065 - **§377** — THREE HARNESS DEFECTS FOUND IN ONE GATING SESSION, ALL "A CONFIDENT NUMBER ABOUT A SMALLER WORLD" (P31 S69) L32107 - **§378** — ★★★ — THE **SELF-CALLER CAST**: LET A TU KEEP CALLING THE FUNCTION IT IS ABOUT TO DEFINE (P31 S69; byte-proven ov_SC04_010/func_8017D6CC) L32124 - **§379** — ★★★ — **MEM_IN_STRUCT_P**: THE SAME LOAD, WRITTEN AS A STRUCT MEMBER, SCHEDULES WHERE A CAST CANNOT (P31 S69; byte-proven main/func_80021284 220 ins and main/func_8002D904 217 ins, found INDEPENDENTLY by two agents) L32174 - **§380** — ★★★ — **A SECOND SET OF A PSEUDO DISQUALIFIES IT FROM `move_movables`** (P31 S69; main/func_800215F4, 465 ins, closeness 106 → 59 → 39) L32205 - **§381** — THE `insn_count` HOIST THRESHOLD IS A DIAL YOU CAN READ WITH `cc1 -dL` (P31 S69; four independent uses in one wave) L32226 - **§382** — TWO FOLD REASSOCIATIONS THAT NEED THEIR OWN STATEMENT (P31 S69) L32242 - **§383** — TWO TOOLCHAIN FACTS THE PACKS DID NOT CARRY (P31 S69) L32254 - **§384** — ★★★ — A CARVE-CONFIG BANK IS RED UNTIL YOU RE-EXTRACT, AND THAT LOOKS EXACTLY LIKE A FALSE BANK (P31 S69; measured twice, cost one destroyed match) L32269 - **§385** — ★★★ — THE **SCHED2 PRIORITY-DONOR ASM**: closing the "hoisted-invariant vs IV-init preheader swap" class (P31 S69; byte-proven main/func_80038A58, 347 ins, fable escalation 2 → 0) L32319 - **§386** — ★★★ — A BYTE LOAD ON THE **BIV** BASE WAS BORN IN THE COMBINE PASS: SPELL IT AS A SHIFT-MASK, NEVER A DEREF (P31 S69; byte-proven main/func_80020598, 292 ins, escalation 1 → 0) L32345 - **§387** — ★★ — **SPLIT-FOLD DISPATCH CLOBBER**: one switch case needs a reload, another must keep the fold (P31 S69; byte-proven main/func_80030F80, 343 ins, escalation 3 → 0) L32371 - **§388** — ★★★ — THE **-O0 COLOURING ORACLE**: simulate `stupid.c` instead of grinding spellings (P31 S69; main/func_80011380 proved a C-level WALL at 6) L32394 - **§389** — ★★★ — `h_norm` IS BLIND TO INDEXED-GLOBAL RELOCS, SO FREE WORK BECOMES AN INVISIBLE SINGLETON (P31 S69; 31 stubs / 4,811 ins recovered, 8 banked same day) L32419 - **§390** — ★★★ — MINIMUM DISTANCE IS NOT MINIMUM WORK; RANK TWIN CANDIDATES BY EFFORT, AND FILTER LOOKALIKES BY RATIO (P31 S69; byte-proven ov_SC01_077/func_80184D50, banked) L32460 - **§391** — ★★ — A BYTE-ALIGNED STRUCT COPIES IN FOUR INSTRUCTIONS, A WORD-ALIGNED ONE IN TWO (P31 S69) L32498 - **§392** — ★★★ — THE SONNET-WAVE HARVEST (P31 S69): SEVEN TYPE/ORDER LEVERS THAT EACH CLOSED A MATCH L32512 - **§393** — ★★ — THE **BIRTHING BOOST**: A SINGLE-SET LOCAL IS SCHEDULED LAST; GIVE IT A SECOND SET (P31 S69; byte-proven ov_SC02_017/func_8017FCFC) L32553 - **§394** — ★★ — TWO ALIGN-1 ACCESSES IN ONE FUNCTION RESERVE A PHANTOM STACK SLOT (P31 S69; ov_SC02_005/func_80180610) L32570 - **§378b** — ★★★ — THE FOUR VARIANTS OF THE DECL BLOCKER, AND THE TWO PLACES §378 DOES **NOT** APPLY (P31 S69, all four measured the same day) L32582 - **§395** — ★★★ — FIVE NARROWING/PLACEMENT LEVERS FROM ONE 91-INSTRUCTION CRACK (P31 S69; byte-proven ov_SC06_018/func_80189E60, warm start 32 off → MATCH 91/91) L32621 - **§314b** — ★★ — TWO ABS-RANGE GUARDS IN ONE FUNCTION NEED **DIFFERENT SPELLINGS** (P31 S69; byte-proven ov_SC04_002/func_8018691C, 111 ins) L32655 - **§322b** — ★★★ — THE CARVE CLASS IS COMPLETABLE, AND EVERY WORKTREE `CARVE-REFUSED` WAS AN INSTRUMENT VERDICT (P31 S69, Fable-3 audit; byte-proven end-to-end) L32677 - **§332b** — ★★★ — THE §332 "WALLS" ARE A PER-OBJECT ASSEMBLER MODE, NOT A C LIMIT — 6 CLOSE AS REAL C (P31 S69, Fable-3) L32712 - **§378c** — ★★ — THE FIFTH DECL-BLOCKER VARIANT: THE DRAFT REDECLARES WHAT THE TU ALREADY OWNS (P31 S69) L32732 - **§396** — ★★★ — SIX LEVERS FROM THE S69 SINGLETON ROUND, INCLUDING ONE THAT ONLY A `COND_EXPR` REACHES (P31 S69) L32747 - **§397** — ★★★ — RE-RUN THE TWIN SCAN AFTER EVERY EXEMPLAR BANK; A CLUSTER SIBLING IS FREE THE MOMENT ITS EXEMPLAR LANDS (P31 S69; ~250k tokens spent proving it the expensive way) L32795 - **§396g** — ★★ — A GUARD LADDER'S RUNGS MUST STAY SYMMETRIC OR `reorg.c` LOSES ITS BRANCH REDIRECT (P31 S69; byte-proven ov_SC03_028/func_80184C90, 92 ins) L32828 - **§398** — ★★★ — `family_remap` CARRIES THE **SOURCE** TU's DECL ENVIRONMENT INTO A DESTINATION THAT ALREADY OWNS THOSE NAMES (P31 S69; measured 3 banked of 22) L32845 - **§398b** — ★★ — NAMING A SUB-EXPRESSION IN A LOCAL CHANGES WHICH PSEUDO SURVIVES; INLINE IT AT BOTH USE SITES (P31 S69; byte-proven ov_SC03_028/func_80183264, 93 ins) L32875 - **§399** — ★★★ — FOUR LEVERS FROM THE FINAL S69 ROUND (P31 S69; each byte-proven, none previously in the cookbook) L32902 - **§400** — ★★ — A BASELINE CHECK THAT CONFLATES "ABSENT EVERYWHERE" WITH "CHANGED UNDER US" SILENTLY DROPS NEW FILES (P31 S69; 8 files, one session) L32942 - **§401** — §401 L32977 - **§402** — §402 L33010 - **§403** — §403 L33034 - **§404** — §404 L33066 - **§405** — ★★★ — THE S70 WAVE HARVEST: 130 agents, 113 MATCH, and the laws they brought back L33102 - **§3-A.** — THE ORACLE HAS A HOLE: match_one cannot see a jump table (§405-A) L33108 - **§3-B.** — THE SCHEDULER DIALS (the dominant residual family this wave) L33118 - **§3-C.** — REGISTER ALLOCATION FROM C, WITHOUT PINS L33134 - **§3-D.** — INTEGRATION IS STILL THE BOTTLENECK, AND THE TU IS THE AUTHORITY L33150 - **§3-E.** — WHAT THE AGENTS REFUTED L33158 - **§406** — ★★ — THE PROLOGUE-WEAVE CLASS IS MECHANICAL AND LARGE (134 open stubs) L33168 - **§407** — ★★ — LATE-WAVE ADDENDA TO §405 (the last agents in) L33190 - **§408** — ★★★ — §406 REFUTED AS A SWEEP: THE SHAPE IS THE FAMILY, THE DISAGREEMENT IS THE DEFECT (P31 S71; 0 MATCH / 14 applied, 0 / 210) L33219 - **§409** — ★★★ — THE S71 JOURNAL-FUELLED WAVE: 100% FIRST-PASS MATCH, AND THE NINE LAWS IT BROUGHT BACK (P31 S71) L33271 - **§3-The** — nine laws this wave produced L33298 - **§410** — ★★★ — COPY THEN ACCUMULATE ON THE COPY: resolving the birthing-boost vs register-allocation dilemma (P31 S71; byte-proven `ov_SC04_015/func_8017EB78`, 98 ins) L33363 - **§411** — ★★★ — THE PACK MUST CARRY THAT FUNCTION'S OWN HISTORY (P31 S71; measured 38/39 vs 124/131) L33396 - **§412** — ★★★ — §323 CARVE BLOCKER 2 WAS A REGEX THAT COULD NOT SEE PAST `__attribute__` (P31 S71) L33434 - **§413** — ★★★ — DIFFICULTY IS THE RESIDUAL CLASS, NOT `nins` — ROUTE THE MODEL TIER OFF HISTORY (P31 S71, Drew) L33483 - **§414** — ★★★ — `parallel_gate` ON `main` IS A FALSE PASS, AND THE RULE WAS ALREADY WRITTEN DOWN (P31 S71) L33520 - **§415** — ★★ — A FILE-SCOPE DECL MAKES gcc-2.7.2 MERGE THE TU'S LATER *BLOCK-SCOPE* EXTERNS INTO IT (P31 S71; byte-proven `ov_SC04_011/func_80180B24`, 215 ins) L33557 - **§416** — ★★ — FOUR LEVERS FROM THE S71 OVERNIGHT LANE, none of which the cookbook held (P31 S71) L33582 - **§417** — ★★★ — A REGISTER PIN CAN BLOCK `jump.c`'s SELECT COLLAPSE, AND UNPINNING THEN EXPOSES A `cse` SKIP-BLOCKS MERGE (P31 S71; byte-proven `ov_SC03_013/func_8017E6F4`, 182 ins) L33615 - **§418** — ★★★ — TWO LOOP-STRUCTURE LEVERS: MAKE THE SECOND INDEX A GIV, AND KEEP A TABLE ADDRESS UNFOLDED (P31 S71; byte-proven `ov_SC04_016/func_8017DF8C`, 184 ins, 32 → 0 in seven compiles) L33642 - **§419** — ★★★ — WHEN A PIN IS IMPOSSIBLE, WIN THE local-alloc DENSITY CONTEST INSTEAD (P31 S71; byte-proven `ov_SC01_000/func_8017DD04`, 297 ins) L33668 - **§420** — ★★★ — A MULTI-CLUSTER SYMBOL REBASE, AND THE BARE-NAME DEDUP THAT HID THREE QUARTERS OF IT (P31 S71; 4 banked in 57 s) L33702 - **§421** — ★★★ — A `la $tN` + `addiu` PAIR CAN BE A **RELOAD** ARTIFACT THAT NO C SPELLING REACHES (P31 S71; byte-proven `md_SC07_003/func_801A293C`, 313 ins, 6 → 0) L33739 - **§422** — ★★ — QImode ARITHMETIC VIA `(u8)(x - K)`, AND `flag ^ 1` NEEDS ITS OWN TEMP (P31 S71; byte-proven `resident/func_800D06E8`, 344 ins) L33770 - **§423** — ★★★ — "MATCH IN ISOLATION + GATE REJECTS + CAUSE NOT DETERMINED" ⇒ GREP THE TU FOR A FILE-SCOPE TYPEDEF THE DRAFT ALSO CARRIES (P31 S71; byte-proven `ov_SC03_092/func_8017FA74`) L33790 - **§424** — ★★★ — EQUAL-PRIORITY STORES COME OUT **REVERSED**: sched1's LUID tie picks the LAST statement first (P31 S71; byte-proven `ov_SC07_006/func_801890FC`, 387 ins) L33817 - **§425** — ★★★ — `sb` ALIASES SCALAR GLOBALS WHILE `sh`/`sw` STRUCT STORES DO NOT, AND TWO MORE ALIAS/BOOST RULES (P31 S71; `md_MAIN_003/func_800CF3E8`, 467 of 469 ins, all four byte-verified from `-dS`/`-dR`/`-dl`/`-dr`) L33839 - **§426** — ★★★ — main's SWITCH FUNCTIONS WERE NEVER A CODEGEN WALL: ONE RODATA CARVE HAD BEEN MISSING SINCE PHASE 7 (P31 S72; 3 of the 11 "PROVEN gate-rejects" banked byte-identical in 14 s) L33865 - **§427** — ★★ — A HASH IS A CORRECTNESS ORACLE WITH ZERO DIAGNOSTIC CONTENT; PRESERVE THE RED ARTIFACT BEFORE ANYTHING REBUILDS OVER IT (P31 S72) L33945 - **§428** — ★★★ — A ZERO-BYTE CROSS-JUMP BARRIER: ADVANCE THE POINTER INSIDE EACH SWITCH ARM (P31 S72; `main/func_80026D64`, 218 ins, MATCH in 2 compiles) L33964 - **§428a** — ★★★ — TWO RESIDUALS THAT MOVE IN OPPOSITE DIRECTIONS UNDER EVERY LEVER USUALLY SHARE ONE CAUSE (P31 S72; `main/func_8001B0D4`, NEAR/53 -> MATCH; **my first answer here was WRONG and is kept below as the refutation**) L34002 - **§429** — ★★★ — EVERY HELD POINTER NEEDS ITS OWN LOCAL, AND A NEGATIVE-DISPLACEMENT BYTE STORE NEEDS ONE OF ITS OWN (P31 S72; `main/CdReadStateMachine`, MATCH 385/385) L34050 - **§430** — ★★★ — A GOTO INTO A LOOP IS FINE; HAND-HOIST THE CONSTANTS IT COSTS YOU (P31 S73 — **this section previously said the OPPOSITE and was wrong; the refutation is kept below**) L34083 - **§431** — ★★★ — SPLITTING A 27,000-LINE TU AT ITS ORIGINAL BOUNDARIES: THE JTBL SPANS TELL YOU WHERE, AND THE COMPILER TELLS YOU WHAT CROSSES (P31 S72; `src/800.c` -> `800.c`/`800_b.c`/`800_c.c`, byte-identical with nothing banked) L34132 - **§433** — ★★★ — ON A SWITCH FUNCTION, CASE SOURCE ORDER IS THE DOMINANT RESIDUAL — AND `match_one` IS BLIND TO IT (P31 S73; 4 of 5 consecutive main MATCHes) L34195 - **§432** — ★★★ — DEFEAT cse's MERGE OF TWO IDENTICAL MASKS BY SPELLING ONE AS A SHIFT PAIR (P31 S72/S73; `main/func_8002DC68`, MATCH 198/198) L34227 - **§434** — ★★★ — TWO SYMBOLS, ONE FRAME: RUN THE FRAME CHECK BEFORE DRAFTING ANYTHING LARGE (P31 S73; `main/SaveLoadRoutine` + `func_8002B0B4`, byte-verified) L34269 - **§435** — ★★★ — AN OVERLAY TU SPLIT IS NEAR-FREE, AND THE GAP TEST IS "IS THIS WORD A CODE ADDRESS", NOT "IS IT ZERO" (P31 S74; four overlays split, all byte-identical, CARVE-BLOCKED class emptied fleet-wide) L34317 - **§436** — ★★★ — TWO TOOLS THAT READ THE WRONG SOURCE OF TRUTH, AND THE SHAPE THEY SHARE (P31 S74; both fixed + negative-controlled) L34374 - **§437** — ★★★ — A WRAPPED TRAILING COMMENT BROKE THE TU PARSER, AND THE §8b DECL LAYER WROTE `extern #define` (P31 S74; five independently-MATCHed jr bodies unblocked, `ov_SC06_029` byte-identical) L34431 - **§438** — ★★★ — THE SAME-ADDRESS LEAD IS NOW SIZE-FILTERED: A HOMONYM IS WORSE THAN NO TWIN (P31 S74; ~12 of ~60 cards carried one) L34538 - **§439** — ★★ — LEVER SET FROM THE S74 WAVE (each entry is one measured crack, not a hypothesis) L34565 - **§440** — ★★★ — A `.rodata` CARVE PIECE BINDS TO A SUBSEG, NOT TO A FUNCTION: EXTEND THE CARVE INSTEAD OF ISOLATING (P31 S74; 6 banked, resident included) L34586 - **§441** — ★★ — THREE MORE INSTRUMENT DEFECTS FROM THE SAME SESSION, ALL OF WHICH BLAME THE SUBJECT (P31 S74) L34627 - **§442** — ★★★ — A RECOVERY RUNG THAT REWRITES A LEGAL CONSTRUCT INTO AN ILLEGAL ONE READS EXACTLY LIKE A CODEGEN WALL (P31 S75; `reconcile_tu`, 344 ins unblocked) L34657 - **§443** — ★★ — A DERIVED DEPENDENCY MUST BE PROVISIONED BY *EVERY* PROVISIONER, AND WE HAVE TWO (P31 S69 → S74, the same defect twice) L34712 - **§444** — ★★★ — A REJECTED GATE DAMAGED COMMITTED CARVE STATE, AND `git status src/` SAID NOTHING WAS WRONG (P31 S75) L34739 - **§445** — ★★ — `make clean BINARY=` IS FLEET-WIDE: THE VARIABLE IS ACCEPTED AND IGNORED (P31 S75) L34788 - **§446** — ★★★ — "RELOC-ONLY REMAP GATED DIFF" IS A CARVE VERDICT, NOT A CODEGEN VERDICT, UNTIL YOU DIFF THE `.text` (P31 S75; 4 siblings + 1 latent, ~1,300 ins) L34817 - **§447** — ★★★ — THE BIGGEST FUNCTION LEFT IS A CARVE, AND THE VERDICT TOOL SAID "DECLARATIONS" (P31 S75; SaveLoadRoutine, 1,165 ins) L34864 - **§448** — ★★★ — ASSEMBLY POSING AS C: 154 GAME FUNCTIONS THE REPORTS COUNTED AS DONE (P31 S75) L34908 - **§449** — ★★★ — FOUR COMPILER DIALS FROM THE S75 REDRAFT WAVE (two byte-gate CONFIRMED, and one correction to §439) L34951 - **§3-A.** — `reg_n_sets` IS A ONE-LINE SCHEDULING DIAL — `t = t + 1;` vs `*p = t + 1` (CONFIRMED: `ov_SC03_105:func_80180ABC`, 257 ins, banked from closeness 250) L34956 - **§3-B.** — A SINGLE-SET LOCAL'S VALUE IS VISIBLE AT A SWITCH JOIN, AND THAT ERASES A ZERO-EXTENSION (CONFIRMED: `ov_SC03_105:func_801806F8`, 241 ins, banked from closeness 235) L34985 - **§3-C.** — CORRECTION TO §439 — `sll 16; srl 16` LANDS *AFTER* THE CALL, AND WORKS FOR A KNOWN CONSTANT L35000 - **§3-D.** — AN OFFLINE JTBL-RODATA PLACEMENT AUDIT (CLAIMED: `md_MAIN_034:func_800CB00C` — did NOT bank, and that is the point) L35018 - **§450** — ★★★ — REGENERATING A TARGET `.s` FOR A FUNCTION THAT IS NO LONGER A STUB (P31 S75; 146 functions unblocked, two silent defects caught by ONE cross-check) L35031 - **§451** — ★★★ — YOUR EVIDENCE HAS MORE THAN ONE SOURCE, AND THE ONE YOU QUERY IS PROBABLY THE WORSE ONE (P31 S75; 37 functions reclassified, 122 ins banked from one word) L35072 - **§452** — ★★★ — NOT EVERY VERBATIM BODY IS UNDECOMPILED WORK, AND §448'S HEADLINE OVERSTATED IT (P31 S75; 10-function burst, 0 banks, and the negative result is the finding) L35110 - **ADDENDUM** — to §265 (P31 S76) — THE PROSE GUARD DID NOT HOLD. IT IS A GATE REFUSAL NOW. L35147 - **CORRECTION** — to §182/§188 (P31 S76) — THE EPILOGUE "WALL" IS AN ORACLE ARTIFACT IN THE REORDER ISLAND L35192 - **§460** — `-dS` PRINTS THE SCHEDULER'S READY LIST WITH PRIORITIES. STOP INFERRING IT FROM RTL ORDER. L35232 - **§461** — LAUNDERING AN INVARIANT CAN BE THE DEFECT, AND "RESIDUAL A" DOES NOT GENERALIZE PAST ONE BINARY OP L35264 - **§465** — THE ASPSX SLOT-HOP: A GAP OUR REORDER SUBSTITUTE CANNOT CLOSE (`main:func_8005F830`) L35294 - **§466** — `main` (509 ins, -O0): ADDRESS CONTEXT EMITS `mult` INDEX-FIRST L35320 - **§464** — FOUR VOLATILE/BARRIER LEVERS FROM `main:func_8005DE78` (141 ins → MATCH) L35338 - **§467** — GLOBAL-ALLOC TIES BREAK ON *DECLARATION* ORDER, AND A COPIED CLOBBER LIST IS A DEFECT L35366 - **§468** — THE `%lo`-FOLD EXTENDS TO STORES ONLY VIA `extern Struct SYM[]`, AND MASKING HID THE OPERAND ORDER L35390 - **§469** — THE `MEM_IN_STRUCT_P` ALIAS UNLOCK (and §463's spill law, independently confirmed) L35424 - **§470** — FOUR CSE/SCHED LEVERS FROM `main:func_800301C8` (170 ins, 133 → 18) L35447 - **§471** — A LAUNDER'S REAL COST IS AN ALLOCNO, AND `$t0` IS RELOAD'S (`main:func_80032A74`, 408 → 12) L35475 - **§472** — 🔴 §148-A's HOIST THRESHOLD IS 29, NOT 58, WHEN THE LOOP CONTAINS A CALL L35499 - **§473** — 🔴 §265's "HANDWRITTEN" VERDICT FOR `ov_SC07_002:func_8017DC80` IS REFUTED (324 → 89) L35529 - **§474** — A *PROVED* C-LEVEL FLOOR: `split_tree` + `stupid.c` (-O0), from `main:func_80011380` L35554 - **§475** — THE `"memory"` FENCE AS A *CSE INVALIDATOR*, AND `(b*3)<<3` INSTEAD OF `b*24` L35591 - **§476** — 🔴 A HARD-REGISTER PIN DESTROYS TWO THINGS COMBINE AND SCHED1 NEED (`func_800226C0`, 670 ins → MATCH) L35622 - **§462** — FOUR LEVERS FROM `main:func_80024054` (91 ins, 74/53/32 → 4) L35650 - **§463** — 🔴 SPILL SLOTS ARE 8 BYTES, AND THE §41b "LOAD ABOVE THE PROLOGUE" WALL IS REFUTED L35687 - **§477** — ★★★ — THE `self_decl_tu` CLASS IS A SOLVED, MECHANICAL LANE: 16 DRAFTS, 16 BANKS (P31 S77) L35723 - **§478** — 🔴 — A VERBATIM DRAFT IS THE STRONGEST *FALSE* SIGNAL YOUR SCOPING TOOL CAN EMIT L35797 - **§479** — ★★★ — WHERE THE PERMUTER ACTUALLY PAYS: A MEASURED YIELD CURVE (P31 S77, 8 candidates) L35817 - **§480** — 🔴 — A STATIC BLOCKER CLASS THAT THE REAL PIPELINE ALREADY REMOVES IS A PHANTOM L35896 - **§481** — ★★★ — `conflicting types` IS A SAME-SCOPE ERROR; ACROSS SCOPES IT IS ONLY A WARNING (P31 S77, `main:func_8001FC08`, 400 ins) L35911 - **§482** — ★★★ — TWO INDEPENDENT RE-TIES, ORDERED: WHEN ONE BARRIER FIXES ONE RESIDUAL AND CREATES THE OTHER (P31 S77, `main:func_8006252C`, 30 ins → MATCH) L35942 - **§483** — ★★★ — THE S77w WAVE HARVEST: SIX LEVERS, FOUR FROM BANKED (BYTE-PROVEN) BODIES L35977 - **§484** — ★★★ — "NO SINGLE NOLOAD BASE" IS NOT "UNLINKABLE": ASK WHETHER THE `.bss` OFFSETS ARE DISJOINT (P31 S77) L36033 - **§485** — ★★★ — THE PLACEMENT MAP WAS PARSING A PRETTY-PRINTER: 25 PsyQ OBJECTS WERE INVISIBLE, NOT ABSENT (P31 S77) L36079 - **§486** — ★★★ — CARVING AN `-O0` ISLAND IN **main**: FIVE COUPLED PIECES, AND THE TWO THAT ANNOUNCE THEMSELVES (P31 S77) L36115 - **§487** — ★★★ — THE PSX LOADER'S PER-VERSION SIGNATURE SETS ARE A FREE PROVENANCE ORACLE: main's "WALL" BAND IS LIBPAD 4.2.1 (P31 S78) L36160 - **§488** — ★★ — THE "GAME CODE" GAPS BETWEEN LIBRARY BLOCKS WERE LIBRARY OBJECTS: 13 SUBSEGS → LINKED, EXACT-TILED, ZERO TOKENS (P31 S78) L36212 - **§489** — ★★★ — SCATTERED `.bss` IS A PARTITION PROBLEM: SPLIT THE SECTION INTO PER-BASE PIECES AT LINK-PREPARE (P31 S78 #4; closes the §9.1 exclusion class 3/3) L36252 - **§490** — ★★★ — THE "WALL" BAND WAS A LIBRARY VERSION AWAY: LIBPAD 4.2.1 + LIBAPI 4.2 FOUND, THE WHOLE 0x8005CE18–0x8005FC68 BAND + THE APICARD REGION LINKED FROM REAL OBJECTS (P31 S79 #13/#5) L36310 - **§491** — ★★ — THE MECHANICAL LEFTOVERS (P31 S79 #6): A PHANTOM STUB, TWO JTBL TWINS, ONE EXACT CLONE — AND THREE TOOL GAPS THE BANKS EXPOSED L36356 - **§492** — ★★ — "C-PLUMBING" WAS THREE DIFFERENT THINGS (P31 S79 #7): A RAW SPLICE THE GATE'S LADDER BROKE, TWO -O0 BODIES THE CHECKER COMPILED AT -O2, AND TWO DRAFTS THAT BELONGED TO OTHER OVERLAYS L36402 - **§493** — ★★ — THE PERMUTER ROUTE END-TO-END, AND THE THREE PLUMBING STEPS BETWEEN A SCORE-0 WINNER AND THE MAIN GATE (P31 S79 #8) L36430 - **§494** — ★★★ — TEN BANKS FROM ONE-AGENT-PER-FUNCTION DRAFTING (P31 S79/S80 #9): THE IDIOMS, THE PLUMBING, AND THE THREE WAYS AN AGENT'S "MATCH" WAS NOT ONE L36490 - **§495** — ★★★ — THE VERBATIM END-STATE (P31 S80 #10): TWO DEF-SIDE DECLARATION WALLS, A "BANK" THAT WAS THE ASSEMBLY, AND A GATE THAT DROPPED A BANK ON EXIT 0 L36630 - **§496** — ★★ — A CARRIED-TYPE TEST THAT ASSUMES THE OVERLAY INCLUDE SET SILENTLY DROPS A RESIDENT TYPEDEF (P32 T1a; byte-proven, resident `func_800D128C` isolation) L36671 - **§497** — ★ — A BODILESS `typedef struct Tag Alias;` DEFINES THE ALIAS, NOT THE TAG: THE CARRIER'S FALSE "CONFLICTING BODIES" REFUSAL (P32 T1b; ov_SC02_017 `func_80186C64` isolation) L36705 - **§498** — ★★★ — A CARVE THAT DROPS THE BINARY'S `--pre` CLAUSE, AND A GATE THAT IGNORED THE EXTRACT'S EXIT CODE: HOW A BYTE-CORRECT DRAFT WAS BOOKED "DIFF" (P32 T1a; resident `func_800D128C` BANKED 243 ins after two instrument fixes) L36728 - **§499** — ★★★ — A NEVER-ONBOARDED PAYLOAD PINS ITS OWN BASE STATICALLY, AND THE FIRST BUILD CANNOT (P32 T2a/T2b: the parked five onboarded, 213 → 218 binaries, 0 UNCLAIMED) L36758 - **§500** — ★★★ — THE T3 WAVE HARVEST (P32 T3, 2026-09-05): 31 one-agent-per-function drafters → 20 MATCH / 9 NEAR / 2 FAIL, every closer, two NEW named gcc mechanisms, and the wave-process defects L36799 - **§501** — ★★★ — A LEVER THAT MEASURES WORSE MAY BE A CASCADE: READ THE `.loop` DUMP FOR THE DESIRABILITY FLIP BEFORE DISCARDING IT (P32 T4b, `main:func_800391D4`, a pinned wall banked by a Fable agent) L37157 - **§453** — ★★★ — ONE SOURCE PER UNIQUE FUNCTION: THE INCLUDE-AT-SITE SHARE, ITS GATE, AND THE FOUR DEFECTS THAT SHAPED IT (P35 S94–S96) L37561 - **§454** — Rung D and rung R: taking a lever off a body that still has to compile to the same bytes (Phase 36 T6) L37610 - **§454a** — the recipe rung's real shape: replication, not discovery (S99, measured) L37677 - **§455** — Rung G, the guided search: score the object, classify the residual, compose the moves (Phase 36 S101) L37692 - **§456** — The lever-removal move catalog: what the S104 agents found under ~205 pins, barriers and launders (Phase 36 T7) L37765 - **§457** — The lever-removal move catalog, S105: eight TU batches, 42 of 42 at 0 (Phase 36 T7) L37814 - **§458** — Struct spelling is a per-access dial: the T2 probe's byte evidence (Phase 37 S106) L37886 --- ## L-number → section (if you cited `§1234` and the grep failed, it was a LINE number) Index rows carry a `L…` anchor = the section's line in `docs/matching-cookbook.md`. Notes routinely quote that as a section id. This table resolves it. Grep bait: `L-number`, `line number cited as section`, `§ grep failed`. | L | section | title | |---|---|---| | L30 | §3-How | to use this | | L39 | §1 | Idiom catalog (asm pattern → C that produces it) | | L41 | §3-I1 | Unsigned range check: `(x - lo) < (hi-lo)` → `addiu`+`sltiu` | | L47 | §3-I2 | Byte mask forces `andi` even after `lbu` | | L54 | §3-I3 | Division by a constant → magic multiply | | L60 | §3-I4 | Runtime (variable) division → `divu` + zero-check `break` (NEEDS `--expand-div`) | | L69 | §2 | Writing matching C (what makes gcc emit X) | | L71 | §3-T1 | Loop pointer: top-of-body for `addu` induction, not constant-folded `addiu` | | L78 | §3-T2 | Source statement order drives instruction scheduling | | L85 | §3-T3 | Types | | L90 | §3-T4 | Branch polarity: invert the source condition to flip gcc's chosen branch | | L107 | §3 | When a diff is pure scheduling → decomp-permuter (harness built, Phase 6) | | L128 | §3a | Escalation TIER above the permuter — web-research the compiler internals (HIGH VALUE, prov | | L147 | §3b | §31-directed permuter mutation — bias the search over the class's levers (Phase 24 T5) | | L190 | §4 | Flag/toolchain gotchas | | L199 | §5 | Known hard-residual classes (instruction-identical, one byte-exact blocker) | | L211 | §5a | Cross-jump tail-merge — gcc collapses two byte-identical blocks the original kept separate | | L265 | §6 | Per-module optimization mixing — the -O0 boot module (Phase 7) | | L273 | Detecting | the opt level (do this first) | | L307 | Build | mechanism — per-file opt override (splat resegmentation) | | L322 | §7 | PsyQ SDK types & symbols (the library-call prerequisite) | | L339 | §8 | rodata island (compiler jump tables) — the `.data→.rodata→.data` sandwich (Phase 7) | | L361 | §8a | rodata island in a flat OVERLAY — the tail sandwich, per matched jr-function (Phase 26 — P | | L399 | §8a-pad | a trailing `.word 0x00000000` under a jtbl dlabel is `.align` PAD, not an entry (Phase 26 | | L423 | §8b | MULTI-jtbl per overlay — the `ld_interleave --order` sandwich + the same-subseg cases (Pha | | L456 | §8c | Splitting a TU means rebuilding its DECLARATION ENVIRONMENT, not moving text (Phase 26 ses | | L502 | §8d | Templating a body INTO a TU must not CHANGE its declaration environment — demote the carri | | L549 | §8e | The jtbl ALIGNMENT LAW + the pad-spec filter — multi-table .rodata spans (Phase 29, byte-p | | L632 | §9 | Link real PsyQ library objects byte-exact (Phase 7 — GO proven) | | L661 | §9.1 | Generalised per-object linker — `tools/psyq_link.py` (+ `psyq_link_lib.py`), 18/18 libcd b | | L685 | §9.2 | Wire a library region into the build with NOLOAD — no data carving (`tools/psyq_link_regio | | L704 | §9.3 | Make it the build: resegment + swap + resolve (`tools/psyq_integrate.py`, libcd DONE) | | L731 | §9.4 | Integrating a SECOND library (libgs block 6 after libcd) — multi-library gotchas | | L759 | §9.5 | Integrating a WHOLE multi-block library in one call (full libgs — Phase 7 session G) | | L785 | §9.6 | Scaling library linking to the whole EXE (Phase 8 — 8 libs linked, 20%→50% byte-identical) | | L821 | §9.7 | Binary-agnostic toolchain refactor (Phase 9) — the reusable pattern for Gen2 | | L854 | §10 | Closing the regalloc/scheduling hard tail by hand (LZSS, Phase 7 session F — the full clos | | L863 | §3-The | clean object-level metric (use this, not the permuter score, for jtbl/rodata functions) | | L875 | Residual | A — commutative `|`/`&`/`+` result lands in the wrong source-operand register | | L886 | Residual | B — a `return ` materialised late / merged instead of distributed per-site | | L927 | §11 | Cross-binary dedup & code-sharing (Phase 11 — "one match unlocks many") | | L945 | §3-The | mechanism: game-code dedup is SOURCE-LEVEL, not an object swap (R-D1, the key lesson) | | L978 | Per-binary | toolchain provenance (R24) | | L983 | §12 | Ultracode harvest — parallel-draft + byte-gate at scale (Phase 12, resident: 1.4%→71.7% in | | L1075 | §13 | Add a location overlay — the canonical runbook (Phase 13; the Phase-15 fleet recipe) | | L1081 | One-command | onboarding — `tools/new_overlay.sh ` | | L1090 | §3-The | flat-blob overlay config (what the template encodes) | | L1098 | §3-THE | NON-4-ALIGNED-OVERLAY GOTCHA (≈75% of the fleet; fixed in the template + Makefile, automat | | L1110 | §3-The | A→B→C per-overlay workflow | | L1127 | Dedup-credit | (§11) for overlays — two shapes | | L1139 | Fleet | build — `make build-all` / `make check-all` | | L1145 | §14 | Propagate a matched function across the fleet — `tools/dedup_propagate.py` (Phase 15) | | L1192 | §14a | The fleet bulk run (Phase 15 — `--auto-from`, 947 REAL → 74,527; 3.82% → 22.14% in one pas | | L1215 | §14b | Harvesting the UNMATCHED shared core — the match_one wall (Phase 15) | | L1236 | §14c | Callee-signature-aware harvest — breaking the extern-type-conflict wall (Phase 15, T6) | | L1292 | §14d | Deterministic recovery beats agent waves on the hard tail (Phase 15, the final-session fin | | L1321 | §14e | Two hard-tail dead-ends (Phase 15 — documented so they aren't re-attempted) | | L1336 | §15 | Struct-heavy shared-core pipeline (Phase 16) — empirical determinations (S0) | | L1358 | §16 | Guided hand-matching the struct-heavy core (Phase 17 — beats the §15 brute-force) | | L1399 | §17 | The compiler-quirk wall — the matching TOOLKIT (Phase 18; gcc-2.7.2 source + byte-gated) | | L1407 | Register-allocation | ORDER (call-crossing $s0/$s1 swap) → FORCE it with register pins (byte-proven) | | L1427 | §3-The | STEERABLE idioms (byte-confirmed this phase) | | L1444 | §3-The | pipeline gotcha — match_one ≠ the gate; `sig_unify` is MANDATORY | | L1451 | §3-The | LOOSE-TYPING wall is real for narrow params (Phase 16, reconfirmed) | | L1458 | Strategic | conclusion — the match-% lever post-research | | L1468 | §17a | The TOOLKIT at WAVE scale (Phase-18 Step-3b/Step-1 — measured) + the pipeline-integration | | L1542 | §18 | Per-file `-O0` split inside an overlay/blob (Phase 19 T1) | | L1620 | §19 | Scaling the toolkit waves — the recovery PIPELINE + the propagation CAP (Phase 19 T3) | | L1655 | §20 | The wave-at-scale GATE CAP + residual-class verdicts (Phase 20) | | L1660 | §3-THE | CAP: at this tail the match_one→gate gap is the LOOSE-TYPING CALL-GRAPH wall (not cheap pl | | L1677 | Diagnostic | lesson: a failed in-TU build leaves a STALE `.o` | | L1683 | §3-NEW | / CONFIRMED residual classes (this session) | | L1704 | Operational | gotchas (cost real time) | | L1712 | What | WORKED — the Phase-20 reusable wins | | L1724 | Phase-20 | RESOLUTION — `tools/cast_call_sites.py` BUILT + the cap re-diagnosed (R14, byte-proven) | | L1775 | §21 | wave-distilled idioms (Phase 21) | | L2028 | §22 | DEF-side loose-typing recovery + grinder blacklist (Phase 21) | | L2055 | §23 | Giant `func_80153E00` cracked (scalar-data CAST); BUT most giant drafts are STALE+INCOMPLE | | L2108 | §24 | The `ov_SC01_077_a.c` split-file vein: split-aware propagation, but loose-typing-gated (Ph | | L2155 | §25 | The "schedule" class is mostly COALESCING (pin-crackable), not scheduling; + the gate two- | | L2167 | §3-The | crack: the residual was gcc-2.7.2 COPY-COALESCING, not the instruction scheduler | | L2183 | §3-The | genuine scheduler — `rank_for_schedule` (sched.c), for when it IS scheduling | | L2192 | §3-The | genuine schedule WALLS (do NOT re-grind — stub) | | L2200 | §3-h | _exact OVER-COUNTS ×134 — verify shareability before crediting a class's "reach-134" (R14) | | L2212 | §3-The | gate two-stage — sig_unify is a FALLBACK, not unconditional (`gate_stage.py`, §19 folded i | | L2222 | §26 | The cheap close=0 recovery lever is EXHAUSTED; `idiom_loop --assess` was DOUBLY inflated ( | | L2228 | §3-The | `_a` close=0 recovery banks 0/20 — same def-side wall as MAIN (0/40) | | L2256 | Where | this leaves the reach-134 tail (cont.6 option-3, now CONFIRMED byte-backed) | | L2282 | §27 | Giant matching recipe (Phase 21 cont.7b — validated on func_80176D94, 152 ins) | | L2348 | §28 | Banking a "close=0 gate-rejected" giant: the canonical-extern recovery (Phase 22 T1, byte- | | L2368 | §28a | decomp.wiki GCC patterns worth trying on BFM giants (decomp.wiki/compilers/GCC, raw at dec | | L2375 | §28c | The close=0 recovery is NOT fully exhausted (§26 corrected, R14); + the dedup_propagate re | | L2380 | §28b | The struct-walled close=0 giant (§28 case #3) IS bankable: the engine_types.h type-lift (P | | L2389 | §28d | The "macro-extern-injection" lever: freeing reach-134 inline matches dedup_propagate skips | | L2397 | §29 | Reasoning-model (GLM5.2) DEF-side reconciliation idioms + the wall's hard limit (Phase 23 | | L2405 | §30 | Fable5Max cracks §20/§10 "unsteerable" from the gcc SOURCE: store-vs-load is a `/s` aliasi | | L2417 | §30a | §30 generalizes to the FULL near-miss backlog (via STANDARD Opus agents, not just Fable5) | | L2425 | §31 | THE gcc-2.7.2 CODEGEN MAP: pass → residual → C-lever catalog (Phase 23; 4 Fable5 agents re | | L2445 | §32 | The region-a CAMERA-GIANT idiom set: struct-base hoisting + 4 sibling levers (Phase 24 T7, | | L2457 | §33 | Automating the giant decl-reconcile: `tools/reconcile_decls.py` (the DATA analog of §20's | | L2477 | §34 | The `func_80138ED0` giant crack: gcc-2.7.2's **3-qty sort bug** + the **zero-byte asm allo | | L2494 | §35 | The region-a sibling-giant harvest: difficulty ≠ $s-reg count (it's global-array hoisting) | | L2510 | §36 | Fable5 giant-crack levers (Phase 24 T7 Fable5 batch; accumulates as each lands — full RTL | | L2529 | §37 | The T7 §G giant endgame: 6/8 cracked, meta-laws + transferable levers (Phase 24, 2026-07-0 | | L2550 | §38 | The WHALE `func_80144B9C` (770 ins): the -O0 struct-assign memcpy idiom + the -O0 reach-13 | | L2560 | §39 | The ×1→×134 giant-endgame: propagate a matched **-O2** giant via the NATIVE DEFINE-macro p | | L2576 | §40 | Structural families: the MECHANICAL symbol-remap (crack one exemplar → remap the rest, ~0 | | L2624 | §40a | The DECL-RECONCILE pass (type-lift so remapped drafts compile in the sibling TU) (Phase 25 | | L2657 | §40b | The reloc-tracker blind spot: the indexed-global idiom that hid the "reach-1 tail" (Phase | | L2699 | §40c | The h_seq per-sibling reconcile: templating a reconcile-class crack ×134 (Phase 26 Task 8, | | L2716 | §31-triage | R17 applies to CODEGEN residuals, never to a compile ERROR (Phase 26 session 7, Drew asked | | L2733 | §41 | The DEF-SIDE canonical-sig wall: mechanically banking a drafted giant past `conflicting ty | | L2792 | §41a | v3.1: the def-side wall was ~71% TOOL-shaped — the five measured defects + the laws that d | | L2847 | §41b | T7 execution: the object-only probe OVER-counts BANKABLE by two link/rodata classes (Phase | | L2877 | §41c | T7-M2: the ×134 def-side-wall sweep via per-sibling RE-reconcile (Phase 25, 2026-07-10; 4, | | L2896 | §41b | addendum — M4 "reconcile_decls" tier is ALSO a probe over-count: 0/8 mechanical (Phase 25 | | L2916 | §41d | `void`→`s32` is NOT always byte-neutral: gate the RAW draft FIRST (Phase 26 session 6, byt | | L2938 | §42 | The F-band ≤28 regalloc crack wave: register-pin/DENSITY levers beat the permuter (Phase 2 | | L2991 | §42a | addendum — wave 2 (residuals + 29-100 band): iso-MATCH ≠ real-TU bank, the memcpy→struct-a | | L3032 | §42b | addendum — wave 3 (Max, 2026-07-10c): THE STALE-OBJECT GATE TRAP + the read-global `&`-cas | | L3071 | §42c | addendum — wave 3 (Max orchestrator + CORRECTED Ultracode fan-out, 2026-07-10c): the real- | | L3120 | §42d | addendum — wave 4 (rtu_match fan-out over the mapped frontier, 2026-07-10c): 24/26 MATCH, | | L3159 | §42e | propagating a CRACK ×134: the def-finder bug + the byte-drift residual (the "remap-fail" m | | L3233 | §43 | The K&R s16-param definition DISSOLVES the "narrow-param wall" for by-value register args | | L3280 | §44 | The Phase-25 cheap-Opus giant batch: 5 structural levers + the §43 extension (2026-07-11, | | L3336 | §45 | The flagship `func_80133CD4` crack (399 ins ×134): the merged-variable permutation-breaker | | L3351 | §46 | The `func_80178D40` crack (890 ins ×134, the heaviest core in the game): four LOOP-STRUCTU | | L3403 | §47 | The live-length SLIDER: splitting a global.c allocno-priority TIE with one zero-byte asm ( | | L3443 | §48 | The 12-core jr crack wave: the ALLOCNO-PRICING dials and the EBB rule (Phase 26 session 8, | | L3451 | §3-A. | ALLOCNO-PRICING DIALS — move a value into the register you want, byte-neutrally | | L3482 | §3-A4 | SINK THE CONSUMER CALL INTO THE ARMS (the inverse of A1; `func_8016AB6C`, byte-proven) | | L3506 | §3-B. | THE EBB RULE — the general form of §46-L2 | | L3522 | §3-C. | TYPE- AND SHAPE-DRIVEN CODEGEN (the C type literally selects the addressing mode) | | L3546 | §3-D. | THE CROSS-JUMP RATCHET (the sharpest new trap — `func_80131340` L-C) | | L3556 | §3-E. | Meta | | L3565 | §49 | The LUID DIAL: a zero-byte SCHEDULING dial (the sched.c analogue of §47) — `func_8017A4AC` | | L3614 | §50 | Refinements that BOUND §47/§48 (from the `func_80135EB0` wall, 21→6; Phase 26 session 8) | | L3668 | §51 | TOOLING INTEGRITY: the silent skip, and how to hunt it | | L3674 | §51a | The bug class | | L3690 | §51b | Why the byte-gate cannot save you | | L3700 | §51c | THE METHOD (do not audit by reading the regex) | | L3715 | §51d | THE LAWS | | L3777 | §51e | The false-wall pipeline (why this is not just hygiene) | | L3793 | §51f | Checklist for any new corpus-scanning tool | | L3807 | §51g | When the thing you are scanning has a GRAMMAR, parse the grammar (`tools/cdecl.py`) | | L3951 | §52 | The WALKER-FAMILY skeleton: 6 regalloc-order levers + a deeply-characterized intrinsic wal | | L3962 | §3-The | 6 levers (the "walker-family skeleton" — apply to the 12 siblings; levers 1-5 retire ~half | | L3983 | §3-Why | the wall is (probably) intrinsic | | L3997 | §52a | The regalloc sibling wave: new levers + two new wall classes (cheap-Opus applying §52, 202 | | L4035 | §52b | Sibling wave 2: more de-pin levers, a third wall class, and the match_one→whole-binary gap | | L4066 | §53 | SWEEP A FAMILY WITH THE TOOL ITS EXEMPLAR NEEDED: the jr/switch carve, and how omitting it | | L4071 | §3-The | case | | L4083 | §3-Why | 0/8 was structural, and predictable from two words | | L4100 | §3-The | rule | | L4116 | §3-The | meta-lesson (R35, and why this one is expensive) | | L4131 | §55 | Core-crack wave levers + the GATE-ORCHESTRATION law (Phase 29 T3, 13-agent ultracode wave, | | L4136 | §55a | New byte-proven levers (each from a banked or near draft) | | L4160 | §55b | THE GATE-ORCHESTRATION LAW (3 traps, ~3.5h lost; all recovered, 0 data lost) | | L4179 | §55c | Sizing the propagate: a TARGETED propagate is ~4 min/core, and "it's slow" was a BROKEN-TR | | L4201 | §54 | `--fix-def-sig`: the member's CANONICAL DECLARATION is a build step too (tiny-IMM mega-poo | | L4225 | §56 | Banking a hand-drafted GIANT into its exemplar TU: self-contained draft vs live-TU decls, | | L4270 | §56b | PROPAGATING an h_seq giant: the exemplar's externs MUST be fleet-canonical, not the draft' | | L4294 | §57 | The SELF-decl normalize: the sibling's OWN caller declares the templated fn divergently (t | | L4341 | §57a | Two NSD corrections + the SURGICAL-ONLY law + the honest broad-sweep yield (Phase 29, 2026 | | L4381 | §58 | match_one MATCH ≠ BANK: the four blind spots + the crack-wave reconcile-before-bank law (P | | L4415 | §59 | Three h_seq sweep-residual classes match_one/the-exemplar-bank don't reveal (Phase 29 crac | | L4445 | §60 | Classify the residual, don't rank it: the deterministic residual→class classifier and what | | L4469 | What | it measured — the whole open backlog, byte-grounded | | L4488 | §3-Two | corollaries worth remembering | | L4501 | §3-The | parallel-probe race this surfaced | | L4510 | §60a | What the first DIRECTED grinder run exposed (Phase 29 Task-13B, 2026-07-21) | | L4548 | §60b | The plateau autopsy's verdict: a `partial` drift is a WRONG DRAFT, not a missing transform | | L4583 | §61 | Task 14: the gate ladder's missing stage is the ARITY pre-pass, and it is TU-side not draf | | L4644 | §61a | The Task-5 wave: 11/12 MATCH, 0 banked — three DISTINCT integration walls, each now named | | L4696 | §61b | The jtbl gate stage: built, and the ORDERING law it exposed (Phase 29 Task-14 stage 4, 202 | | L4751 | §61c | The jtbl bank is INCREMENTALLY valid and CLEAN-INVALID (Phase 29, 2026-07-21) — the blocki | | L4817 | §61d | The undo was eating the tree: two tools, one defect, invisible to the byte-gate (Phase 29, | | L4871 | §62 | The jtbl RECONCILE must also follow the carve: the post-carve draft reconcile (`harvest_ve | | L4915 | §63 | The fresh-138 DEF-SIDE blocker: fix the HEADER decl, not the draft (`fix_header_decl.py`, | | L4946 | §64 | The §20 type-lift's three laws: fold the tagged typedef, check VISIBILITY, and strip only | | L4995 | §64a | VARIANT types: UNIQUIFY the camps, do not reconcile them (`uniquify_type.py`, Phase 29 SES | | L5038 | §63 | UPDATE (Phase 29 SESSION-14) — `fix_header_decl`'s "SAFE" verdict is FLEET-BLIND; it MUST | | L5053 | §65 | The stranded-draft recovery: BLAST-RADIUS TIERS, and the per-overlay de-macroize that refu | | L5068 | §65a | The blast-radius taxonomy (makes §61's law structural instead of remembered) | | L5083 | §65b | The escape: de-macroize the instantiation, don't touch the shared header | | L5119 | §65c | `rtu_match` MATCH → bank held 13/13 on self-decl, but broke on the FIRST callee-decl case | | L5129 | §65d | Existing-ladder baseline, measured (do this before building a recovery stage) | | L5139 | §65e | Two oracles, and the disagreement is the finding (R34 in practice) | | L5156 | §65f | The de-macroize lever's BOUNDARY, measured: byte-neutral 14 times of 15, and the 15th is w | | L5177 | §65g | Where the cheap levers STOP: the local-type and in-TU-self-decl classes did not yield | | L5197 | §66 | Exercise a banking driver's SUCCESS path before pointing it at a wave: the free re-bank te | | L5219 | §66a | The widest write in a pipeline is the one most likely to be UNDECLARED | | L5238 | §66b | A metric parsed out of another tool's prose goes NULL silently when the label changes | | L5255 | §66c | Before a wave, verify the FUEL exists; an "already attempted" set built from the wrong dir | | L5283 | §66d | The permuter⇄reader loop: alternate a random search with a byte-verified idiom, and let `r | | L5318 | §66d-1 | What transfers between giants is the LOOP, not the PIN | | L5327 | §66d-2 | Two operational sharp edges | | L5338 | §66d-3 | Read the ILS per-cycle SERIES, not its final best: a repeated score and a still-falling on | | L5353 | §67 | The arg-copy PLACEMENT lever: launder a parameter into a fresh pseudo AT the statement whe | | L5443 | §67a | Run the symbol-set guard BEFORE you pay for a gate (`tools/symcheck.py`, Phase 29 SESSION- | | L5476 | §66d-4 | "ILS converged" means converged FOR THAT WEIGHT PROFILE, not a floor (amends §66d-3; Phase | | L5529 | §66d-5 | `residual_class`'s "structural ⇒ permuter CPU is waste" is WRONG for schedule permutations | | L5562 | §68 | A comment-only line halted the extern scan, and the skip label blamed the type cap (Phase | | L5607 | §69 | How to attack a behemoth: map it, don't draft it (Phase 29 SESSION-18, `func_80183814`, 5, | | L5657 | §70 | The giv-init base register: walk the PARAMETER, not a copy of it (Phase 29 SESSION-18, `fu | | L5698 | §71 | Before mapping a giant, look for an already-matched SIBLING beside it (Phase 29 SESSION-18 | | L5764 | §72 | A `register __asm__` pin is a PREFERENCE, not a reservation (Phase 29 SESSION-18, `func_80 | | L5788 | §3-The | honest fix was source-level and cheap | | L5798 | §3-Giv | record order (the §70 family) | | L5804 | What | is left, and what is byte-recorded as SPENT | | L5821 | §73 | A def-side self-decl conflict has TWO axes: RETURN (fleet widen, T2) and PARAMS (casts at | | L5860 | §74 | Auditing a pinned draft: the §72 hazard is CALLER-SAVED pins spanning a call, and only the | | L5904 | §75 | A propagation cap is usually a MINORITY-SPELLING SOURCE OVERLAY, not a wall: census the ca | | L5963 | §75a | The exclusion classes, enumerated with named causes (Phase 29 SESSION-19, the 134-binary ` | | L5986 | §75b | A body's preamble can carry `#define`s, not just `extern`s; extraction lifts only the exte | | L6035 | §75c | Class-B's remedy is the FULL §17a-1 PAIR (decl **and** call-site cast); a decl-only fix mo | | L6066 | §76 | The allocno CLASS (local vs global) is the dominant regalloc lever, and C reaches it ONLY | | L6075 | §3-The | mechanism, with citations | | L6095 | §3-The | attribution primitive (use this before calling anything a scheduling residual) | | L6102 | §3-Two | diagnosis traps this function proved | | L6112 | Practice | Practice | | L6122 | §77 | Every extraction tool carries a NARROW hard-coded preamble set; anything outside it silent | | L6162 | §3-The | CANDIDATE gate and the REAL gate need DIFFERENT preambles — keep the difference out of the | | L6212 | §78 | A LENGTH drift can be a register grant in disguise; and `fold` never leaves a literal firs | | L6218 | §3-The | drift was an allocation decision, not missing code | | L6249 | §3-The | economics | | L6258 | §79 | For a 0-callee giant, fingerprint by DATA symbols (§71 cannot fire); and the STACK-SLOT OR | | L6264 | §71 | has a blind spot, and this is it | | L6278 | §3-NEW | LEVER — the frame layout reads back the original declaration order | | L6289 | §76 | confirmed at scale, and a pin nuance | | L6298 | §3-The | residual, and the honest read | | L6308 | §80 | A do-not-re-buy entry is scoped to its BASE, not to the function; and the pin's hidden cos | | L6329 | §3-The | pin's hidden cost, with the citation | | L6340 | §3-The | flagged "#1 move" LOST — and why the failure is informative | | L6350 | §78 | 's attribution primitive, run and reproduced | | L6356 | Cold-start | economics, now complete | | L6361 | §81 | Banking a jr (jump-table) function: the 3-step carve chain, and why `match_one` cannot see | | L6392 | §3-The | defect this exposed: a shared type that is present but invisible | | L6408 | §82 | Two source-shape oracles from behemoth #6: a duplicated `addiu $aN,$sp,K` across a `jal` m | | L6413 | §3-1. | The inlined-helper signature | | L6427 | §3-2. | Scalar vs aggregate decides *when* the slot is allocated | | L6439 | Also | reproduced on this function | | L6443 | §3-And | the banking footnote (§75a class A, one line) | | L6450 | §83 | The parameterised-repeat law, the spill-area trap, and why a per-case edit cannot move a p | | L6457 | §83a | READ THE HEADLINE NUMBER CORRECTLY: a LENGTH drift makes `match_one`'s count meaningless | | L6465 | §83b | THE LEVER: find the parameterised REPEAT before decoding case-by-case | | L6483 | §83c | TRAP: a "dead local" in a prior draft may be gcc's OWN spill area | | L6490 | §83d | CSE's quantity budget is WHOLE-FUNCTION, so a local rewrite cannot fix a local symptom | | L6504 | §83e | §80 vindicated again, on the same day it was written | | L6510 | §84 | The DERIVED-OFFSET remap bug: a hand-computed literal that encodes the DISTANCE between tw | | L6515 | §3-The | construct | | L6533 | §3-Why | it survived every candidate gate | | L6548 | §3-THE | FIX IS MECHANICAL — the tool already holds the answer | | L6556 | Scope | , measured (do not over-generalise — §80) | | L6566 | §3-Two | ladder lessons banked with it | | L6577 | §85 | The RETURN-axis fleet widen is ALL-OR-NOTHING: widening the shared header alone guarantees | | L6582 | §3-The | conflict | | L6590 | §3-THE | FAILURE MODE — widening only `engine_core.h` is worse than not starting | | L6597 | §3-The | precondition, and how to check it in one grep | | L6605 | §3-Do | the WHOLE axis in one edit, then R22 once | | L6614 | Reading | , for the next person | | L6627 | §86 | Pinned-exemplar templatability is a PER-FAMILY property, not a per-member rate; and the §4 | | L6629 | §3-The | guard refuses a class that largely works | | L6638 | §3-THE | LAW: all-or-nothing PER FAMILY | | L6665 | §3-Why | the two live families differ from the three dead ones — the open question | | L6671 | §87 | `match_one` COMPILES but never LINKS, so an unresolvable data symbol reads as MATCH; and s | | L6687 | §3-The | blindness ladder, now complete — FOUR classes `match_one` cannot see | | L6697 | Consequence | for the backlog ledger | | L6704 | §3-The | cheap discriminator, before spending a sweep | | L6712 | §88 | `cross_jump` will not merge a common suffix containing a CALL; and §78 is scoped to ORDERE | | L6717 | §88a | repeated CALL-shaped blocks are left UNMERGED; call-free tails are merged for you | | L6724 | §88b | the `slti` literal-position law (extends §78 to comparisons) | | L6739 | §88d | BANKING ORDER: run the §81 carve chain BEFORE banking, never after | | L6746 | §88e | a wrong diagnosis, refuted properly (the model for how to treat an inherited lever) | | L6756 | §88f | the missing rung: a RELOCATION gate between `match_one` and the binary | | L6764 | §89 | Two throughput rules the project already had written down and was not following (Phase 29 | | L6770 | §89a | MEASURE the write set; do not assert its tier (`tools/blast_radius.py`) | | L6783 | §89b | the parallel gate farm existed; the family path could not reach it (`tools/sweep_parallel. | | L6796 | §3-The | standing sequence | | L6804 | §90 | Five tool-integrity laws from one session, each of which changed an answer (Phase 29 SESSI | | L6810 | §90a | A comparison tool MUST share its reference oracle's index space, exactly | | L6827 | §90b | "Byte-neutral" is not "wanted": undo on the SUCCESS path too | | L6837 | §90c | A library-callable function must FAIL CLOSED on an unconfigured module | | L6848 | §90d | Do not measure a live wave's drafts (§87 in real time) | | L6856 | §90e | An agent's CONCLUSION and its EVIDENCE fail independently — re-derive the premise, not the | | L6880 | §91 | A structure-TRANSFER is only valid where the structure corresponds: the `--like` role trap | | L6909 | §3-The | three-hypothesis trail, because two of them were wrong and the wrongness is instructive | | L6929 | §92 | Conforming a DECLARATION: pointer changes are caller-neutral, scalar-WIDTH changes are not | | L6960 | §93 | `set -o pipefail` attributes a pipeline failure to the LAST stage, not the failing one (Ph | | L6985 | §94 | A family sweep's `0/N` is a TYPE-CARRY failure until proven otherwise: lift the exemplar's | | L7021 | §95 | `reconcile_tu` dropped the SIBLING declarators of a multi-symbol `extern` line (Phase 29 S | | L7056 | §96 | The same rewrite, one shape down: `reconcile_tu` matched statements to lines by TEXT, so e | | L7104 | §97 | The gate's own tree hygiene: a refused carve, an unchecked recovery, and a snapshot that c | | L7151 | §98 | `conform_decls` had three defects, and only the third needed R22 to find (Phase 29 SESSION | | L7205 | §99 | The narrow-param wall is a DEF-side problem with a ZERO-blast-radius fix: convert the defi | | L7228 | §3-Two | `reconcile_tu` bugs found underneath, one introduced while fixing the other | | L7254 | §100 | Prefer the DRAFT-LOCAL fix: a type only one function uses belongs in its BODY, not in a sh | | L7283 | §101 | The STALE DEFAULT class: a guard whose cause was removed is a silent skip wearing a safety | | L7310 | §102 | A PLUMBING verdict can MASK a DIFF; and K&R is not always a codegen change (Phase 29 SESSI | | L7345 | §103 | A FILE-scope `extern` in a shared overlay TU is a GLOBAL constraint on every LATER functio | | L7433 | §104 | Two silent-skip defects in one scan: match on MASKED text, emit from the ORIGINAL (Phase 2 | | L7469 | §105 | A gate's revert must survive an EXCEPTION, not just a failure (Phase 29 T53, `jtbl_family_ | | L7508 | §106 | Persist the MEASUREMENT, derive the POLICY: a stored route let a stale file out-vote the l | | L7551 | §107 | A lever wired into ONE gate path is a lever most families cannot reach (Phase 29 T56, `fun | | L7589 | §108 | Diagnosing a family `0/N`: the four causes, and the third opt-in lever (Phase 29 T59) | | L7638 | §109 | Conforming a definition to a shared header: fix the NAMES, then check the RETURN precondit | | L7678 | §110 | A unit must define exactly ONE function, and "ends in `;`" does not tell you which line de | | L7719 | §111 | The distinct-code metric is not noisy: a family pays it only if its members are byte-VARIA | | L7768 | §112 | A macro-scoped declaration only collides where the macro is INSTANTIATED (Phase 29 T67/T69 | | L7814 | §113 | An ARITY blocker only exists if the macro CALLS the function; an address-taken use has no | | L7846 | §114 | The THIRD decl axis: a CALLEE the draft declares differently from the target TU (Phase 29 | | L7888 | §115 | A `func_XXXXXXXX` predicate rots by design: the same name-form assumption in THREE places | | L7915 | §116 | Optimization level is a property of the FILE, not the function: read a family 0/N against | | L7933 | §3-The | fix moves the DEFINITION, not the stub — and here is why the obvious shortcut fails | | L7964 | §117 | Spell the sibling's symbol from the SIBLING's address, not the exemplar's kind (Phase 29 T | | L7991 | §3-Why | it survived so long: a MASKED oracle will MATCH a wrong symbol | | L8001 | §118 | Ordinal (positional) immediate resolution: compare C tokens to the DIFFERING asm uses (Pha | | L8035 | §119 | Two levers on the SAME axis, opposite directions: test the off-diagonal (Phase 29 T89) | | L8064 | §120 | Uniquify draft-defined TYPE names; and check which of N staging sites you actually patched | | L8075 | §3-Do | NOT "strip the duplicate typedef" — it breaks the extern that uses it | | L8081 | §3-The | wiring trap that cost two attempts | | L8099 | §121 | Synthesise externs for macro-DEFINED callees from the macro's own definition head (Phase 2 | | L8122 | §122 | GATE RAW BEFORE TRANSFORMING; the undo belongs to the WRITER, as a per-edit journal (P30 T | | L8154 | §123 | PROPAGATE A FAMILY WITH THE TOOL ITS TIER NEEDS: `dedup_propagate` is h_exact-only; its re | | L8191 | §124 | A "not matched" verdict can mean the definition is there under a DIFFERENT C NAME: the asm | | L8236 | §124a | a family sweep's `0 matched-exemplar families` may be a FILTER, not a wall | | L8244 | §125 | Split the CARVE from the BODY before calling a jr residue a wall — and measure it by SHA f | | L8251 | §3-The | method (keep this) | | L8261 | §3-The | instrument rules that make its answer trustworthy (this is where I failed) | | L8276 | §3-The | corrected results (each SHA-verified, from a clean tree, restore re-verified) | | L8286 | §3-Two | further notes worth keeping | | L8295 | §3-The | meta-lesson | | L8302 | §126 | The carve-within-a-carve: an ADDRESS RANGE is not an OPTIMIZATION REGION (P30 T2, byte-pro | | L8317 | §3-The | finding: opt level is per FILE, so the file's contents must be opt-HOMOGENEOUS | | L8328 | §3-The | instrument trap that hid it (and it is §124's shape again) | | L8337 | §3-The | mechanics | | L8355 | §126a | a bare `except: continue` around a coverage-asserting oracle re-creates the silent skip (P | | L8387 | §127 | The `-O0` regime: the CONSTANT-OFFSET FOLD, and why `-O0` needs its own idiom set (P30 T3 | | L8394 | §3-The | idiom they kept re-deriving: the constant-offset fold | | L8405 | §3-The | rest of the `-O0` regime (write PLAIN C, and mean it) | | L8415 | §127a | §71 (sibling-first) is the strongest `-O0` lever, and it beats the index | | L8424 | §127b | the knowledge was in a SOURCE COMMENT, not the cookbook | | L8431 | §128 | A raw NUL in C source makes grep SILENTLY SKIP the file (P30 S28, 137 files) | | L8463 | §128a | a negative control must corrupt a SCRATCH COPY, never the tracked file | | L8475 | §129 | Post-carve, `rtu_match`/`match_one` COUNT THE JUMP TABLE AS INSTRUCTIONS; and a carve must | | L8479 | §129a | the target instruction count is INFLATED after a carve | | L8500 | §129b | never commit a carve whose owner is still a stub (it strands the carve) | | L8515 | §3-The | real blocker underneath, for the record | | L8524 | §130 | An INCREMENTAL build can report BYTE-IDENTICAL for a change the CLEAN build cannot even LI | | L8561 | §3-The | diagnostic ladder that finally located it (reusable) | | L8571 | §131 | The jtbl OVER-SPAN: `sltiu N` is ground truth in BOTH directions, and the zero-word rule o | | L8608 | §132 | The `JR-PAIR-IN-ONE-O0-OBJECT` "wall" was TWO instrument defects: a merged-double span the | | L8617 | Defect | 1 — a pre-§8e MERGED DOUBLE is not a single-table predecessor | | L8635 | Defect | 2 — `as` writes a corpse and nothing deletes it | | L8645 | §3-The | fingerprint, and the 30-second ladder that found it | | L8666 | §3-The | transferable rule | | L8673 | §132a | `--like` is for a sibling with NO record; against one that HAS a record it over-derives (P | | L8698 | §132b | When the span's already-matched owner is ITSELF multi-switch: `--span-rel` (P30 S1, `func_ | | L8723 | §133 | The DEFAULT-FILTER class: three times in one session, a tool silently answered a narrower | | L8749 | §134 | MULTI-LINE BLINDNESS: one root cause, four faces, in `family_remap`'s preamble scanner (P3 | | L8803 | §135 | Six byte-verified gcc-2.7.2 idioms from the P30 S6f-h waves (and the two-lane wave shape t | | L8808 | §3-The | codegen idioms | | L8842 | §3-The | integration idioms (these decide whether a byte-correct draft BANKS) | | L8857 | §3-The | wave shape that produced these | | L8875 | §136 | The LOCAL-VARIABLE lever: how many C locals, at what scope (P30 wave 4a, 25 byte-verified | | L8891 | §3-The | splitting/merging rules (each closed a residual, byte-gated) | | L8917 | §3-The | type-form rules | | L8950 | §3-The | scheduling rules (refining §135-2 and §135-4) | | L8979 | §3-The | declaration surface (integration, not codegen) | | L8991 | Wave | economics (measured, for the next batch's sizing) | | L9014 | §136a | Blocker capture: classify on the OUTPUT, never on the exit status | | L9071 | §136b | A prior wave's "genuine byte-DIFF" verdict is NOT reliable evidence (4 of 4 refuted) | | L9107 | §136c | SIBLING-FIRST is a DERIVATION shortcut, not just a conflict fix (the fastest route in a fa | | L9132 | §136d | Four gcc-2.7.2 levers the redraft lane found (each closed a residual no other lever moved) | | L9176 | §136e | §136c's PRECONDITION, and two more symptom keys (wave 4b batch 3) | | L9224 | §136f | Two declaration sub-cases the reconcile lane surfaced (lane now 15/15 lifetime) | | L9249 | §136g | When the index points at the WRONG lever: two byte-refuted routings (func_801863B4) | | L9278 | §136h | CORRECTION: the zero-crack pool does NOT "refill with cheap work" (my error, byte-measured | | L9308 | §136i | The drafter model LADDER: Haiku → **Sonnet** → Opus → Fable5 (Drew, 2026-08-03) | | L9340 | §136j | The failure MIX flips with function size (measured across four bands, one session) | | L9382 | §137 | REGALLOC-PERM is a TWO-COMPILE ARITHMETIC PROBLEM, not a permuter job | | L9424 | §137a | A gate verdict has a TIMESTAMP; re-check it against the draft's mtime | | L9453 | §138 | The propagation lanes: a gate refusal is a DECLARATION, and which lever you owe depends on | | L9459 | §3-The | triage, cheapest first | | L9511 | §3-The | DEFINITION-side alias is the only escape when the fleet canon disagrees on a promoting par | | L9519 | Rank | the lane by measured concentration, not by class count | | L9528 | THREE | carry variants hide in one "CARRY-FIXABLE" bucket — and they need different fixes | | L9564 | §134 | again, in a second tool — and the waiter rule corrected | | L9584 | STEP | 0 of sibling-first: grep `src/` for a distinctive LITERAL from the `.s` | | L9603 | Reconciling | a gate-refused draft: which way you edit depends on WHERE the TU's decl is | | L9632 | §139 | A GATE THAT GREPS FOR VERDICTS MUST ASSERT 1:1 ACCOUNTING; and a `--src` filter must not s | | L9664 | §3-The | generalisation — three corollaries worth more than the bug | | L9681 | §3-And | the inverse-lookup trap, same session | | L9698 | §140 | A METRIC IS NOT A MEASUREMENT UNTIL IT IS REPRODUCIBLE FROM THE COMMITTED TREE (P30 S1e: a | | L9718 | §3-The | three-line proof (do this before diagnosing any metric movement) | | L9727 | §3-The | two instrument defects it exposed | | L9747 | §3-The | same swallow, twice more, in the integration spine | | L9758 | §3-Two | wrong mechanisms I chased first, and why they were wrong | | L9778 | §141 | The §134 class is CLOSED: every line-shape decision now routes through `cdecl._mask` (P30 | | L9816 | §142 | An open stub whose `h_exact` class is MATCHED elsewhere is FREE. Propagate the body; do no | | L9828 | §3-The | measurement (do this before any wave; it is ~20 lines and needs no builds) | | L9843 | §3-The | trap that hid it — SAME FUNCTION, TWO ROUTES, ONLY ONE IS FREE | | L9852 | Route | selection (why `--addr` sometimes says "nothing changed") | | L9860 | §3-And | the report-vs-bytes lesson attached to it | | L9872 | §143 | `cast_call_sites` read a RETURN STATEMENT as a prototype and deleted it. A 0/39 sweep beca | | L9926 | §144 | THE LITERAL'S SPELLING PICKS THE IMMEDIATE ENCODING (P30 S40 wave 1, `func_801822E0`) | | L9965 | §145 | Three loop/combine levers from the S40 wave-2 drafters (16/16 match_one) | | L10015 | §146 | RE-MEASURE A WALL BEFORE YOU RESPECT IT. Both "permanent" giants fell to drafts already on | | L10026 | §3-Why | a correct draft can read as an intrinsic wall | | L10038 | Then | propagation returned 0/137 TWICE — both times a missing TYPE | | L10047 | §3-Two | errors of mine, both instructive | | L10061 | §3-The | rule | | L10074 | §147 | The three-stratum FRAME LAW, and four "stop searching" verdicts (P30 S42, `func_8017C294`, | | L10079 | §3-A. | The frame has THREE strata, and stratum 3 is unreachable from C | | L10095 | §3-B. | A `?:` on MEMORY operands costs ~16 bytes of invisible frame; on REGISTER operands, zero | | L10106 | §3-C. | Inner-block declaration does NOT delay slot allocation — BYTE-REFUTED | | L10111 | §3-D. | A lone `$t8`/`$t9` in the target is RELOAD SCRATCH — reproduce the spill, don't pin the re | | L10121 | §3-E. | A `qty_compare` TIE is not spelling-reachable — recognise it and stop | | L10133 | Consequence | for the family (a real scheduling decision) | | L10183 | §148 | The loop.c hoisting THRESHOLD is arithmetic you can compute, and the `?:` clamp that folds | | L10189 | §3-A. | `move_movables` hoists iff `threshold × savings × lifetime ≥ insn_count` — and you can rea | | L10208 | §3-B. | `(v < 0x40) ? v : 0x3F` is folded to `MIN_EXPR` and expands to the WRONG SHAPE | | L10222 | §3-C. | A zero-byte ALLOCNO-PRIORITY slider | | L10232 | §3-D. | Reproduce the original's BUGS verbatim | | L10281 | §149 | Four instrument defects in one session, and the two questions they were hiding (P30 S43) | | L10287 | §3-A. | A prep step that returns its input on failure is indistinguishable from a search that foun | | L10305 | §3-B. | Same address + same name ≠ same body — and the ledger keys on address | | L10321 | §3-C. | `make: *** [...] Error N` is a summary, never a diagnosis | | L10331 | §3-D. | "Cheap fuel" that was never probed: 0 of 31 templatable | | L10350 | §150 | A register ROTATION across symmetric blocks is VARIABLE-IDENTITY evidence, not an allocato | | L10357 | §3-The | fix | | L10367 | §3-The | method that found it (this is the transferable part) | | L10383 | §3-Two | corrections to the record | | L10394 | Diagnostic | order (adopt this) | | L10405 | §151 | THE GHOST WEDGE: when a load-before-store transposition is unreachable by ANY statement or | | L10410 | §3-The | mechanism (read from cc1's own `-dR` trace, not inferred) | | L10425 | §3-The | lever — a zero-emission insn that absorbs the blocked tick | | L10433 | §3-The | two fallouts, and how to close them (both measured, in order) | | L10445 | When | to reach for it | | L10456 | §152 | BYTE SIZE is the family key that name- and h_seq-grouping both miss (P30 S43, the 0xECC fa | | L10461 | §3-The | finding | | L10470 | §3-The | key | | L10481 | §3-Two | cautions that must travel with this technique | | L10492 | §3-The | companion defect (open) | | L10504 | §153 | THE ADDRESS-REMATERIALISATION LAUNDER: a third zero-emission asm lever (P30 S43, `func_801 | | L10512 | Symptom | Symptom | | L10517 | Mechanism | (gcc source + RTL dumps, not inferred) | | L10524 | What | does NOT work (14 byte-measured probes) | | L10530 | §3-The | cure — a fresh launder per site, each in its own block | | L10540 | Companion | levers from the same function | | L10560 | §154 | Reading a disc payload: the module-id word, static base derivation, and "type 1 = uncompre | | L10565 | §3-A. | Payload word0 is a global MODULE ID; code starts after the header | | L10574 | §3-B. | Two static base-derivation methods that must AGREE (use both) | | L10586 | §3-C. | PAC type 1 = the same payload class as type 4, just NOT compressed | | L10596 | §155 | hi/lo literal scanning MUST track base registers (S45) | | L10605 | §155a | the same failure class, one level up: SHAPE-blind table scanning (S45 p5) | | L10625 | §155b | check the TYPE your oracle returns before comparing against it (S45 p5) | | L10648 | §155c | the ZERO-REFERENCE trap: gcc splits a global-array address across the `lui` and the LOAD ( | | L10678 | §156 | an ORPHANED reconcile poisons the fleet: `dedup_propagate`'s kept edit (S45 p6/p7) | | L10725 | §157 | the cheap-tier size cliff, measured (S45 p6) | | L10750 | §158 | The RANGE-EXTENDER: a fourth zero-emission asm lever completes the allocno toolkit (P30 S4 | | L10759 | Symptom | Symptom | | L10765 | §3-Why | the fork is chained (gcc source, validated insn-by-insn against -dS/-dR dumps) | | L10778 | §3-The | method (dump-arithmetic first, then place — no probing) | | L10795 | Bonus | facts worth keeping | | L10810 | §156 | THE PREFERENCE-DONOR MERGE: cross-region variable reuse is what fills a0-a3, and a call-ar | | L10866 | §159 | THE DECLARATION AXIS: conform to byte-truth, and make every guard state its COVERAGE (P30 | | L10931 | §160 | THE REACH-15 WAVE HARVEST: an align-1 block move, and the instrument that called it a wall | | L11007 | §161 | THE RETRY-WAVE HARVEST: a jump table indexed from zero, and two allocator traps (P30 S47) | | L11052 | §162 | S48 WAVE-1 HARVEST (P30, 2026-08-11): the reach-ordered sibling campaign's first 12 target | | L11079 | §162a | SHARPENS *(sharpens §161a, §131, §8a-pad, §129a)* | | L11108 | §162b | SHARPENS *(sharpens §48-A3, §156, §150, §76)* | | L11137 | §162d | SHARPENS *(sharpens §31, §21, §30, §55a)* | | L11179 | §162e | NEW | | L11181 | §162 | THE LICM PAIR: what makes an address a movable AT ALL, and why the preheader order is the | | L11247 | §162f | SHARPENS *(sharpens §42d, §41d, §73, §10)* | | L11302 | §162g | NEW | | L11304 | §162 | CROSS-JUMP DIRECTION: the surviving copy is always the LATER one, so a BACKWARD `j` into a | | L11338 | §162h | SHARPENS *(sharpens §88, §88a, §50-B, §8)* | | L11340 | §162 | The cross-jump "CALL veto" is a COUNT law, not a CALL law (BOUNDS §88a; P30 S48, `func_801 | | L11406 | §162i | SHARPENS *(sharpens §135, §21, §42, §32)* | | L11431 | §162j | SHARPENS *(sharpens §25, §136d-1, §48-B, §46-L2)* | | L11460 | §162k | SHARPENS *(sharpens §1-I2, §12, §160d, §21)* | | L11523 | §162l | SHARPENS *(sharpens §48-B, §48-C1, §20, §21)* | | L11578 | §162m | SHARPENS *(sharpens §36, §158, §148, §153)* | | L11580 | §158a | THE FIFTH LEVER IS NOT AN ASM: `do { } while (0)` is a REGION ref-multiplier you MINT (P30 | | L11587 | §3-The | law | | L11598 | Size | it before you write it (§158 step 1-2, applied) | | L11619 | §3-The | wrap BOUNDARY is the dial — and it is indiscriminate | | L11625 | §3-Not | a pure dial | | L11629 | DIAGNOSTIC | TELL — two faces, one law | | L11651 | §162n | NEW | | L11684 | §162o | SHARPENS *(sharpens §158, §136-1, §136-6, §79)* | | L11743 | §162p | SHARPENS *(sharpens §48-B, §46-L2, §156, §136d-1)* | | L11760 | §162q | SHARPENS *(sharpens §30, §30a, §135-2, §136-13)* | | L11798 | §163 | S48 WAVES 2-3 HARVEST (P30, 2026-08-11/12): the five that were byte-probed and are actiona | | L11866 | §163z | THE UNVETTED REMAINDER (do not cite as law; each needs a dedupe pass) | | L11886 | §164 | S48 §163z SKEPTIC PASS (P30, 2026-08-12): 190 claims vetted, 82 banked | | L12338 | §16Xy | SHARPENS *(sharpens §136d-3, §37 /s-DEP LATTICE, §135-2, §136-13, §162q)* | | L12340 | §3-The | `/s` drop clause is in ALL THREE dependence predicates, so the FIXED-ADDRESS STORE is what | | L13674 | §164z | REFUTED CLAIMS: do NOT re-derive these | | L13740 | §165 | S48 WAVE-4 HARVEST (P30, 2026-08-12): banked the same day the wave landed | | L14398 | §16Z | SHARPENS *(sharpens §37 "the /s-DEP LATTICE", §136-13, §136-14, §16Xy, §162q, `gcc-2.7.2- | | L14400 | §3-The | ADDRESS-CLASS TABLE: which load/store pairs even REACH the `/s` clause (P30 S48 wave 4, `f | | L14916 | §165z | REFUTED THIS WAVE: do NOT re-derive | | L14960 | §166 | THE DESTINATION-TU ORACLE (P30 S48): the seven-attempt bug that was never codegen | | L15016 | §167 | S48 WAVE-5/6 HARVEST (P30, 2026-08-12): the saturation point | | L16212 | §167z | REFUTED IN WAVES 5/6: do NOT re-derive | | L16269 | §168 | THE COUSIN TIER (P30 S49, 2026-08-12): h_seq's exact-hash brittleness, measured — and the | | L16321 | §169 | THE MICRO-ADAPT LANE (P30 S49): edit a proven body, don't crack a new one | | L16368 | §170 | THE A-PROP WORD-DIFF CARD (P30 S49): the lane that had no owner | | L16412 | §171 | THE STALE SEED SYMBOL (P30 S50, 2026-08-13): why §170's 91%→57% was never codegen | | L16478 | §171a | THE MECHANICAL A-PROP DRAFT (P30 S50): 256 members banked with no agent in the loop | | L16528 | §171b | THREE CARRIES THE MECHANICAL DRAFT NEEDS (P30 S50, banking the top-reach families) | | L16561 | §172 | THE ORPHAN-SLOT MECHANISM v2 (P30 S50-Max): the complete frame-residue model for gcc-2.7.2 | | L16615 | §172a | TWO DECOMPILATION TELLS FROM THE SAME DIG (P30 S50-Max) | | L16631 | §172b | THREE MORE TELLS FROM THE GCC READ (P30 S50-Max, banked on Drew's ask) | | L16674 | §173 | THE STORED-PLUMBING RECOVERY RECIPE (P31 T6): symfix-first, per-group isolation, and where | | L16710 | §174 | THE ADAPT-CARD WAVE RECIPE (P31 waves A/B, 2026-08-14): prevention beats recovery | | L16780 | §175 | A CALLER-SAVED REGISTER PIN CAN BE A CORRECTNESS BUG, NOT JUST A SCHEDULING CHOICE (P31 wa | | L16805 | §176a | THE VERIFICATION-LAYER LAWS (P31 overnight, 2026-08-15). What each check can and cannot pr | | L16845 | §176b | BATCH-GATING MECHANICS (P31): what changes when N drafts land in ONE .c | | L16871 | §176d | THE CONFLICT TABLE MUST BE SEEDED FROM THE TU, AND KEYED PER FILE (P31 S52, 2026-08-15) | | L16910 | §176e | SYMBOL IDENTITY IS COMPUTABLE OFFLINE (P31 S52): `tools/reloc_identity.py` | | L16967 | §176f | THE DECLARATION FORM IS A MATCHING LEVER, SO RECONCILE TOWARD THE FORM THE MATCH NEEDS (P3 | | L16998 | §176g | SIZE A WAVE BY INSTRUCTIONS, NOT BY CARDS (P31 S52 — the adopted doctrine) | | L17036 | §176h | THE BATCH-SUBSTITUTION HAZARD MAP (P31 S52): seven holes, three wrong fixes, one law | | L17044 | §3-A. | The seven under-reporting holes (all in `gate_main`, all the same shape) | | L17062 | §3-B. | Typedef handling — the only strategy that survives contact | | L17084 | §3-C. | The limit that remains (recorded, not solved) | | L17091 | §3-C2. | RECONCILE BEFORE THE FIRST GATE — a parked draft gets HARDER to bank, not easier | | L17114 | §3-D. | The measured cost shape, and what to build next | | L17132 | §176i | WHAT A STATIC PRE-GATE CHECK CAN AND CANNOT PROVE (P31 S52, wave Q) | | L17164 | §176j | STOPPING A WAVE MID-FLIGHT COSTS THE IN-FLIGHT TAIL (and how much is recoverable) | | L17190 | §176j-2 | THE REPAIR PASS, MEASURED (do this instead of resuming) | | L17210 | §176k | TWO SELECTOR BUGS THAT SILENTLY SHRINK A WAVE | | L17227 | §177 | 🔴 THE EPILOGUE RETURN-DELAY SLOT IS DECIDED BY YOUR SAVED-REGISTER SET, NOT BY SCHEDULING | | L17284 | §178 | SIX LEVERS MINED FROM THE WAVE-P JOURNALS (P31 S52), each byte-proven and source-cited | | L17296 | §3-A. | THE `$0`-ADD OPAQUE COPY defeats `make_regs_eqv` (func_80033398, 93 ins) | | L17312 | §3-B. | A `return ` IS A PRIORITY-1 HARD-REG SET THE SCHEDULER PLACES FIRST (func_8001BE30, | | L17325 | §3-C. | SINGLE-SET TEMPS GET THE BIRTHING BOOST (func_8001D3FC, 196 ins) | | L17333 | §3-D. | A NARROW TYPE BLOCKS COPY ELISION (func_8001D3FC — new idiom) | | L17340 | §3-E. | THE ZERO-OFFSET ALIAS HOLE (func_80037028, 71 ins) | | L17349 | §3-F. | `MEM_IN_STRUCT_P` ASYMMETRY IN `true_dependence` (func_80037144, 124 ins) | | L17355 | §3-G. | TWO MODELLING TRAPS THAT COST THESE AGENTS SWEEPS OF HUNDREDS OF COMPILES | | L17371 | §179 | IDIOMS MINED FROM THE WAVE P/Q JOURNALS (P31 S52, harvest pass) | | L17381 | §179-A | 🔴 A LOOP-WALKED POINTER **PARAMETER** HANDS ITS ARGUMENT REGISTER TO THE GIV (9 byte-proof | | L17430 | §179-B | 🔴 THE HAND-WRITTEN-ASM TRANSCRIPTION CHECKLIST (maspsx + masked_diff; 10 byte-proofs) | | L17480 | §179-C | 🔴 A FUNCTION WITH NO EPILOGUE (falls into a sibling's shared tail) MUST BE FILE-SCOPE `__a | | L17531 | §179-D | `gte_stflg` MUST CLOBBER `"$12"` OR THE WHOLE TEMP FILE ROTATES BY ONE | | L17551 | §179-E | A `>2*MAX_MOVE_BYTES` BLOCK COPY IS A **STRUCT ASSIGNMENT**, NOT A HAND LOOP | | L17577 | §179-F | PINNING A LOOP-WALKED POINTER IS A TOTAL OFF-SWITCH FOR STRENGTH REDUCTION | | L17600 | §179-G | 🟡 A PIN CAN **CREATE** A COMBINE `LOG_LINK` AND DELETE AN `andi` (sixth RC-5 channel, n=1) | | L17627 | §179-H | A MID-BODY `.global LABEL` PAIR SLICES A BYTE-COMPARABLE FRAGMENT OUT OF A LARGER ROUTINE | | L17650 | Considered | and NOT banked | | L17667 | §176c | MAIN (SLUS_007.26) CANNOT BE GATED INCREMENTALLY | | L17678 | §176 | SEVEN LEVERS FROM THE P31 OVERNIGHT WAVES (2026-08-15): statement order, false regalloc, a | | L17684 | §176-A | "SCHEDULE / DELAY-SLOT / LENGTH-DRIFT ±1" ⇒ check STATEMENT ORDER around the call first | | L17708 | §176-B | "REGALLOC-PERM, 1-4 instructions off" ⇒ it is usually NOT register allocation | | L17754 | §176-C | 🔴 WALL REFUTATION: a hard-register pin CANNOT schedule around a call, because of a genuine | | L17780 | §176-D | CSE-class levers used in reverse (two sharpenings of §153 and cse_expr §2) | | L17806 | §176-E | Two cheap source spellings, both cc1-probed | | L17828 | §176-F | Misdiagnosis triage: four residual verdicts that were lying | | L17845 | What | is NOT banked here | | L17858 | §180 | THE LEFTOVER-DRAFT HARVEST: RE-VERIFY WHAT YOU ALREADY HAVE BEFORE DRAFTING ANYTHING NEW | | L17887 | §180b | WHAT THE PRE-GATE LADDER ACTUALLY FINDS IN A COLD PILE (the shape of integration debt) | | L17906 | §180c | WHEN A BINARY'S MASS BAND IS SPENT, THE FLEET-WIDE DRAW IS STRICTLY BETTER | | L17924 | §181 | WHAT A WAVE'S GATE ACTUALLY REJECTS (P31 S53, measured on wave R's 45-draft main pile) | | L17925 | Only | ONE of 27 blocked drafts was wrong. The other 26 were correct and unbankable. | | L17980 | §182 | §177's HONEST NEGATIVE: the epilogue lever cracked 4 of 16, and the `800c3` cluster held | | L17994 | §180d | THE `pgrep` BRACKET TRICK PROTECTS THE PATTERN, NOT THE COMMAND LINE | | L18006 | §183 | THE DECLARATION-RECONCILIATION PLAYBOOK (P31 S53, measured on 20 byte-verified drafts) | | L18007 | §3-18 | of 20 reconciled while keeping the match. The two that did not are mechanism, not effort. | | L18077 | §184 | COMMENT-BLINDNESS IS A DEFECT CLASS, NOT A BUG (P31 S53: three tools, one root cause, one | | L18105 | §185 | EDIT THE SIDE THAT IS CHEAP TO VERIFY, AND CHECK A TU RETYPE AT ITS USE SITES | | L18137 | §186 | CROSS-JUMPING RUNS **AFTER** SCHEDULING, SO NO C-LEVEL BARRIER CAN STEER IT | | L18158 | §186b | A NO-SAVE 16-BYTE FRAME IN A LEAF FUNCTION MEANS `s16` LOCALS, NOT A HIDDEN CALL | | L18166 | §186c | WHERE A VALUE IS LOADED DECIDES WHICH ALLOCATOR OWNS IT, AND THEREFORE ITS REGISTER | | L18183 | §187 | 🔴 "SAME SOURCE" IS NOT "SAME OBJECT": THE SDK BUILD AND THE GAME BUILD DISAGREE ON GTE NOP | | L18221 | §188 | 🔴 THE `jr $ra` + `addiu $sp` TAIL IS AN **ASSEMBLER** ARTIFACT, NOT A FRAME SHAPE | | L18271 | §189 | FIVE COMPILER LAWS MINED FROM THE WAVE R/S JOURNALS (P31 S53), each source-cited and re-de | | L18346 | §190 | THREE PRESCRIPTIONS FROM THE SAME HARVEST (weaker evidence than §189, honestly labelled) | | L18383 | §191 | WHAT THIS HARVEST DID **NOT** BANK (4 rejected, 4 narrowed) — recorded so it is not re-der | | L18405 | §192 | THE PRE-GATE LADDER WAS MAIN-ONLY, AND NOBODY COULD SEE IT (P31 S54) | | L18406 | Three | defects in one call path; the overlay slates that carry most of the wave work were being w | | L18470 | §193 | THE WAVE-T HARVEST (P31 S54): 71 index_gap reports -> 9 laws, 5 rejected, 61 already-cover | | L18487 | §193-A | The wave card ships only OPEN-set pointers and discards the atlas's BANKED one — `exemplar | | L18554 | §193-B | A NARROW CAST OF A WIDE PARAMETER COSTS ONE EXTRA INSTRUCTION WHEN ITS STATEMENT SITS AFTE | | L18592 | §193-C | gcc-2.7.2 cross_jump merges the SCHEDULED common SUFFIX only — there is no prefix/head mer | | L18626 | §193-D | A COMPILER-GENERATED ARGUMENT COPY IS A `optimize_reg_copy_1` TRIGGER: when the pointer's | | L18697 | §193-E | A varying-address (pointer) load is re-emitted once per CSE-LIVE INTERVAL, and naming it i | | L18727 | §193-F | §148-A2 — The `threshold -= 3` STAIRCASE: `move_movables` admits a COUNT of invariants, no | | L18791 | §193-G | §164-54's "scope to ≥4 arms" bound is byte-wrong — the dispatch-topology oracle goes live | | L18852 | §193-H | A pointer-derived base load `*(s32*)(p+K)` is uncacheable across ANY memory write or call | | L18892 | §193-I | A DECLARED AGGREGATE LOCAL HAS AN 8-BYTE FRAME STRIDE — CEIL(size,8), NOT size. N ARRAYS T | | L18951 | §193-REJECTED | what this harvest did NOT bank (recorded so it is not re-derived) | | L18978 | §194 | THE WAVE-U HARVEST (P31 S54): 64 index_gap reports -> 14 laws, 5 rejected, 44 already-cove | | L18995 | §194-A | A zero-byte scheduling fence goes AFTER the defining statement to make that computation em | | L19061 | §194-B | A `addu $rA,$rB,$zero` copy feeding ≥2 `sh` stores is a SECOND, 16-BIT-DECLARED local — th | | L19099 | §194-C | A CALLER-SAVED loop counter proves its live range crosses ZERO calls — so it cannot share | | L19162 | §194-D | Declared width of a computed-value local is a sched1 dial (count-neutral) — a replication | | L19203 | §194-E | The wave card's `exemplar` is the TARGET ITSELF on 42/73 wave-U and 36/71 wave-T cards — a | | L19240 | §194-F | `if ((*p = v = f()) == 0)` is an expand-time pseudo SPLITTER (store_expr's `want_value && | | L19313 | §194-G | Reading the integer half of a 16.16 stack aggregate: REGISTER-LIVENESS, not the C spelling | | L19356 | §194-H | §164-29's WAR fence runs again at SCHED2 ON HARD REGISTERS: an in-place `v &= K` before a | | L19394 | §194-I | §16N+2's magic-per-odd-part ladder has exactly one broken row — read the divisor arithmeti | | L19437 | §194-J | Back-to-back identical stores: flow.c's `last_mem_set` deletes the first, and only `volati | | L19489 | §194-K | Blind sched1's alias oracle with a second SET of a pointer pseudo — the first zero-byte, n | | L19567 | §194-L | §88b and §189-E are BOTH half-wrong, but not the way the candidate says: the compare-const | | L19615 | §194-M | A STORE in a CONDITIONAL branch's delay slot proves its C statement DOMINATES the branch — | | L19673 | §194-N | §193-D's C dial is misstated: the lever is a SURVIVING CODE_LABEL (a label with a real inc | | L19733 | §194-REJECTED | what this harvest did NOT bank (recorded so it is not re-derived) | | L19754 | §195 | THE WAVE-V HARVEST (P31 S54): 67 index_gap reports -> 14 laws, 9 rejected, 76 already-cove | | L19768 | §195-A | §167-08's "an $aN READ before the jal is scratch" has a byte-proven FALSE-NEGATIVE class: | | L19839 | §195-B | A CALL_INSN does not start a basic block in gcc-2.7.2 — so a call-crossing temp can be a L | | L19881 | §195-C | A call-argument `%hi/%lo` pair sitting at the block head, far above its `jal`, is a load-d | | L19929 | §195-D | §195 — `masked_diff.mask_for` returns 0 for EVERY `j`/`jal` word, so an internal `j` desti | | L19997 | §195-E | A 0/1 materialised at a JOIN immediately before the controlling `beqz`/`bnez` proves the s | | L20055 | §195-F | fold-const.c:4825 canonicalises a `?:` whose THEN arm is zero (or constant against a non-c | | L20107 | §195-G | §NEW — TWO ARMS CALLING THE SAME CALLEE MERGE INTO ONE `jal` UNLESS EACH ARM'S OWN CODE CO | | L20155 | §195-H | §165-27g — ON A FIXED-SYMBOL GLOBAL, `extern T D[];` + `D[0]` IS THE CSE RELOAD DIAL, AT Z | | L20202 | §195-I | §145(b) AMENDED — the pointer-bump `addiu` is saved by cse, not combine: ANY second SET of | | L20267 | §195-J | GTE / PsyQ op CALL-vs-INLINE is a PER-SITE SOURCE FACT, not a TU style — both forms coexis | | L20311 | §195-K | At -O2 the §18 array-of-struct lever is a FRAME lever, not a length lever — but only when | | L20363 | §195-L | The cse store-re-seed does not cross a JOIN LABEL: per-arm stores + a join read keep the r | | L20415 | §195-M | Frame `vars` is a SEQUENTIAL bump-allocation, not a flat sum: §193-I's CEIL(aggregate,8) t | | L20464 | §195-N | In a call-bearing chain of N≥2 `if (f(...)) return 1;` tests closed by `return 0;`, the LA | | L20511 | §195-REJECTED | what this harvest did NOT bank (recorded so it is not re-derived) | | L20557 | §196 | PUT ON THE CARD WHAT THE TREE ALREADY KNOWS: the fleet's declaration consensus (P31 S54) | | L20612 | §197 | THE WAVE-W HARVEST (P31 S54): 68 index_gap reports -> 4 laws, 3 rejected, 41 already-cover | | L20626 | §197-A | A NARROW SIGNED MEMORY READ FEEDING A CONSTANT `>>` LOSES ITS `lh`, AND THE CURE IS AN ASM | | L20670 | §197-B | A REPEATED COMPARE OF ONE VALUE AGAINST ONE CONSTANT IS DELETED BY cse's `qty_comparison_c | | L20698 | §197-C | Fix A1 (operand order) cannot move a commutative destination whose .greg conflict set alre | | L20743 | §197-REJECTED | §197-REJECTED | | L20758 | §199 | THE WAVE-X HARVEST (P31 S54/S55): 63 index_gap reports -> 7 laws, 2 rejected, 56 already-c | | L20769 | §199-A | §189-A's asm→source inference is byte-FALSE: an interloper between a split constant's `lui | | L20818 | §199-B | A permutation sweep that holds ANY statement fixed is not a sweep: the statement an agent | | L20863 | §199-C | A NEGATIVE CONSTANT MULTIPLY ALWAYS TAKES expmed's negate_variant — but whether you ever S | | L20911 | §199-D | A narrow UNSIGNED value compared in an ordered `if` emits `sltiu`/`sltu`; the only dial is | | L20971 | §199-E | §189-A BOUNDED AND CORRECTED — the discriminator is INSN_PRIORITY, not "is the interloper | | L21030 | §199-F | §164-36b — THE TARGET-HEAD FENCE IS A DELAY-SLOT THREAD SELECTOR, AND IT ONLY FIRES WHEN ` | | L21084 | §199-G | At TWO case nodes the switch-vs-if oracle is not blind — but the tell is ALL tests positiv | | L21139 | §199-REJECTED | §199-REJECTED | | L21150 | §200 | THE ALIAS IS THE UNIVERSAL DECLARATION ESCAPE: stop negotiating with the TU's spelling (P3 | | L21199 | §201 | THE WAVE-Y HARVEST (P31 S55): 67 gap reports -> 5 laws, 8 rejected, 53 already-covered | | L21209 | §201-A | §150-B applies to `decl_prior`'s DEF row: for an overlay-window symbol the banked "definit | | L21241 | §201-B | In a narrowed PLUS/MINUS/AND/IOR/XOR expression the destination pointee is INERT — the sig | | L21321 | §201-C | §X — A CALL'S OWN DELAY SLOT AND THE UPSTREAM CONDITIONAL BRANCH'S SLOT COMPETE FOR ONE IN | | L21372 | §201-D | THE SIGNEDNESS OF A div/mod MAGIC IS DECIDED BY THE STATIC TYPE OF THE DIVIDEND TREE AFTER | | L21419 | §201-E | §194-J-2 — The `last_mem_set` deletion window is measured in SOURCE/expand order, not in t | | L21475 | §201-REJECTED | eight, the session's highest | | L21500 | §202 | THE ALIAS CARRIES A DEFINITION, NOT JUST A DECLARATION: the DEF-SIDE-RETURN wall (P31 S56) | | L21537 | §203 | A DEDUPED TYPEDEF MUST PRECEDE EVERY SPLICE POINT, NOT JUST ITS OWN (P31 S56) | | L21598 | §204 | THE WAVE-Z HARVEST (P31 S56): 82 gap reports -> 5 laws, 16 rejected, 30 already-covered | | L21617 | §204-A | A COMPARE THAT APPEARS BOTH IN A BRANCH'S DELAY SLOT AND AGAIN ON THE FALL-THROUGH IS A JO | | L21692 | §204-B | A LOOP COUNT THAT ARRIVES ON THE STACK IS DECREMENTED IN PLACE: a fresh counter local can | | L21756 | §204-C | WHEN A LOCAL BUFFER'S ADDRESS IS PASSED TO A CALL, ITS SIZE IS A FACT ABOUT THE CALLEE'S B | | L21814 | §204-D | A LOOP-INVARIANT LOAD IS ADMITTED AS A MOVABLE ONLY IF ITS ADDRESS CANNOT TRAP: `local.fie | | L21889 | §204-E | `decl_prior`'s `%hi/%lo` ARM HAS NEVER FIRED: the card's promised GLOBAL-TYPE row is 0 of | | L21941 | §204-CONFIRMED | 30 reports that the index already answered | | L22035 | §204-REJECTED | sixteen, twice the previous record | | L22108 | §205 | THE CHAINED ASSIGNMENT IS ITS OWN SCHEDULING DIAL: `*b = *a = v;` moves an argument copy t | | L22163 | §206 | THE JTBL-CARVE DRAFTING IDIOM: the bounds check is the entry count, and an EMPTY case owns | | L22232 | §207 | THE WAVE ab–ag HARVEST (P31 S58): 278 byte-banked notes → 25 laws, 103 self-reported no-ga | | L22253 | §208 | TWO NAMED LOCALS FOR ONE RELOADED EXPRESSION BUY TWO ALLOCNOS — the naming granularity own | | L22302 | §209 | THE NARROW LOCAL IS A DIAL IN TWO OPPOSITE DIRECTIONS, AND §194-B's "≥2 `sh` STORES" BOUND | | L22379 | §210 | THE SINGLE-BIT MASK IN A BOOLEAN TAIL: `andi K ; sltu $zero,v` vs `srl n ; andi 1` is a ST | | L22425 | §211 | HOIST THE LOOP INIT ABOVE THE DOMINATING GUARD: it fills the guard's delay slot AND flips | | L22484 | §212 | THE WALKING CURSOR IS COUNTABLE: `*wp++` emits one `addiu` PER STORE, `wp[0..2]` emits one | | L22529 | §213 | INDEPENDENT SAME-BASE STORES: THE EMISSION ORDER IS A PERMUTATION OF SOURCE ORDER, AND THE | | L22565 | §214 | THE BANKED TWIN MAY BE A MACRO, A DELETED `.s`, OR A SEMANTIC INVERSE — six ways a ≥0.9 si | | L22611 | §215 | PIN ECONOMY: the twin's pins are NOT part of the shape, and §17's "pin every call-crossing | | L22667 | §216 | DISTINCT ADJACENT SCALARS vs ONE ARRAY: one `lui` per access is the tell, and the array de | | L22707 | §217 | DECODING A CALL'S STACK ARGUMENT SLOTS: `sw` at `0x10`/`0x14`/`0x18` are params 5/6/7 **(s | | L22733 | §218 | A NARROW TYPE AT THE ABI BOUNDARY COSTS AN IN-PLACE `sll/sra` PAIR — on the RETURN as well | | L22762 | §219 | COMPOUND `+=`, FULL ASSIGNMENT, AND AN EXPLICIT TEMP ARE THREE DIFFERENT SCHEDULES OF ONE | | L22789 | §220 | THE PARAMETER ITSELF IS A REGALLOC DIAL: use it directly, prefer `s32` to `void*`, and pla | | L22829 | §221 | A CONSTANT SHARED BY TWO STORES DIES AT THE CALL WHOSE DELAY SLOT REFILLS ITS REGISTER **( | | L22845 | §222 | SWITCH vs IF-CHAIN, PART 3: source arm order IS emission order, a leading EMPTY case buys | | L22887 | §223 | READING A `jal` DELAY SLOT: the value in it was produced BEFORE the call, so it is NEVER t | | L22949 | §224 | CROSS-JUMP: WRITE THE DUPLICATE, AND READ A SHARED DELAY SLOT AS THE MERGE SIGNATURE (P31 | | L22985 | §225 | THREE CONTROL-FLOW SHAPES NO STRUCTURED SPELLING REACHES (P31 S58) | | L23028 | §226 | FRAME PADS: FOUR WAYS §162i1/§2429's DEAD-LOCAL LEVER MISFIRES (P31 S58) | | L23069 | §227 | TYPE THE SOURCE BY THE **LOAD** WIDTH, NOT BY THE STORE WIDTH (P31 S58) | | L23089 | §228 | READING THE DIVIDE, PART N: the off-by-one compare is `% K == 1`, and three more discrimin | | L23126 | §229 | THE ADDRESS IS A VALUE: NAMING IT MOVES THE `lui`/`addiu` PAIR — AND §L14410 SAYS THE OPPO | | L23175 | §230 | THE ANCHOR PROBE: with `%hi`/`%lo` masked, the surviving `addiu` deltas tell you which ass | | L23190 | §231 | TRANSCRIPTION AND SEMANTIC-READ HYGIENE: six ways the listing misleads (P31 S58) | | L23231 | §232 | WHEN THE `jr` DELAY SLOT'S STORE STORES THE RETURN VALUE, TIE THEM WITH ONE PSEUDO **(sing | | L23267 | §30 | addendum (P31 S58) — the `/s` grant closes a SCHEDULING residual and a REGISTER residual w | | L23287 | §194-B | addendum (P31 S58) — BOUND 2 is byte-wrong; see §209 Direction A | | L23293 | §176-B | addendum (P31 S58) — the misdiagnosis direction | | L23299 | §165-40 | addendum (P31 S58) — the barrier goes at the COPY SITE, not inside the region it protects | | L23308 | §164-63 | addendum (P31 S58) — the interposed asm works when the SECOND value is an ordinary assignm | | L23317 | §193-A | / §194-E addendum (P31 S58) — where the twin's body actually lives | | L23323 | §233 | THE WAVE aa–bg HARVEST (P31 S58b): 1,101 byte-banked notes → 24 new laws, 21 addenda, ~700 | | L23367 | §234 | CONSTANT MATERIALISATION: THE STORE LVALUE'S SIGNEDNESS PICKS `addiu` vs `li`/`ori` (P31 S | | L23408 | §235 | THE PHANTOM SYMBOL: A MASKED `MATCH` CAN CARRY A RELOCATION THAT DOES NOT EXIST (P31 S58b) | | L23437 | §236 | THE DECLARATION LAYER IS THE DOMINANT BANK-BLOCKER: NINE WAYS A BYTE-PERFECT BODY FAILS TH | | L23516 | §237 | THE CAST-AT-CALL-SITE DECISION TABLE: WHAT §17a-1 FIXES, WHAT IT CANNOT, AND THE FOUR ESCA | | L23571 | §238 | SAME NAME, DIFFERENT FUNCTION: THE OVERLAY-HOMONYM TRAP (P31 S58b) | | L23611 | §239 | TWO-STATEMENT INTEGER-SPACE MATERIALISATION REORDERS `la` vs `sll`; AND THE PLUS-TREE OPER | | L23645 | §240 | `A + K + B`: WRITE THE CONSTANT **BETWEEN** THE TWO RUNTIME TERMS (P31 S58b) | | L23674 | §241 | THE FOLDED SIGN-EXTEND-AND-SCALE: `sll 16 ; sra (16 − log2 scale)` (P31 S58b) | | L23703 | §242 | `*k` vs `<>n`: EXPRESSION SPELLING OWNS THE LOAD WIDTH AND THE ROUNDI | | L23731 | §243 | SPELL THE CSE BARRIER AS A REASSIGNED POINTER; AND THE `nop`-AFTER-EVERY-`lh` SIGNATURE (P | | L23761 | §244 | `volatile` IS A COUNTING INSTRUMENT, A STORE-ORDER PIN, AND MUST SOMETIMES BE ASYMMETRIC ( | | L23796 | §245 | THE CALL'S ARGUMENT LIST IS A SCHEDULING SLOT (P31 S58b) | | L23843 | §246 | THREE-LIVE-VALUE SCAN LOOPS WANT ADDRESS-FROM-INDEX; AND TWO SYMBOLS CAN SHARE ONE giv (P3 | | L23883 | §247 | TWO BRANCHES TO **ONE** LABEL MEANS THE SOURCE CONDITION IS NEGATED (P31 S58b) | | L23917 | §248 | SPLIT THE LOAD FROM THE ARITHMETIC: A FUSED `g + K` DENIES THE CALLEE-SAVED REGISTER ITS D | | L23941 | §249 | THE SELF-ASSIGN, THE DEAD RE-ASSIGN, AND THE `+ zr` COPY: THREE WAYS TO MAKE A DELETED INS | | L23983 | §250 | `%hi/%lo` vs `lw`: THE EXTERN'S ARRAY-vs-SCALAR SHAPE DECIDES ADDRESS MATERIALISATION (P31 | | L24020 | §251 | IMMEDIATE-SPELLING TRIGGERS: `+= 0xFF`, FULL-WIDTH `~K`, AND THE TWO-OR SPLIT (P31 S58b) | | L24042 | §252 | THE GUARDED PRE-DECREMENT: `(x != 0) && (--x == 0)` (P31 S58b) | | L24063 | §253 | POSTFIX `++` vs `+= 1` PICKS A DIFFERENT SCRATCH REGISTER **(single observation — not yet | | L24077 | §254 | THE DEAD PARAMETER IS A REGISTER-PLACEMENT TOOL **(single observation — not yet cross-conf | | L24089 | §255 | THE EMPTY CASE, PART 2: FOUR TREE SHAPES IT BUYS (P31 S58b) | | L24124 | §256 | GOTOS IN THE TARGET'S BLOCK ORDER REPRODUCE SWITCH PLACEMENT WITHOUT SWITCH'S SIDE EFFECTS | | L24161 | §257 | THE DEAD-END LEDGER (P31 S58b): ELEVEN LEVERS THAT MEASURED NULL OR BACKFIRED | | L24204 | §258 | ADDENDA TO EXISTING SECTIONS (P31 S58b) | | L24209 | §30 | addendum (P31 S58b) — THE ANONYMOUS STRUCT MEMBER REF GRANTS `/s`, AND THAT IS A TWO-FOR-O | | L24230 | §194-B | / §209 addendum (P31 S58b) — TWO MORE INSTANCES, AND THE BOUND IS NOW REFUTED FOUR WAYS | | L24253 | §202 | addendum (P31 S58b) — THE DEF-SIDE ALIAS ALSO CLEARS A RETURN+PARAM DOUBLE CONFLICT | | L24264 | §205 | addendum (P31 S58b) — CHAINED ASSIGNMENT: N≥3 IS INNERMOST-FIRST, AND THE TEXT MIRRORS EMI | | L24281 | §208 | addendum (P31 S58b) — IT SCALES TO SIX SITES, AND IT HAS AN EXACT INVERSE | | L24310 | §210 | addendum (P31 S58b) — THREE CONFIRMED SPELLINGS OF THE BOOLEAN TAIL | | L24326 | §211 | addendum (P31 S58b) — INIT PLACEMENT: FIVE MORE DIALS BEYOND THE GUARD HOIST | | L24369 | §213 | addendum (P31 S58b) — THREE MORE PERMUTATION LAWS FOR INDEPENDENT SAME-BASE STORES | | L24398 | §214 | addendum (P31 S58b) — FOUR MORE WAYS A HIGH-SIMILARITY TWIN LIES | | L24434 | §215 | addendum (P31 S58b) — PIN ECONOMY, PART 2: NINE REFINEMENTS | | L24486 | §217 | CROSS-CONFIRMED (P31 S58b) — AND THE INCOMING HOME SLOT IS THE MIRROR | | L24503 | §220 | addendum (P31 S58b) — THE PARAMETER, NOT A COPY (SEVEN CARDS) | | L24528 | §222 | addendum (P31 S58b) — IF-CHAIN vs SWITCH: THREE MORE DISCRIMINATORS | | L24546 | §223 | addendum (P31 S58b) — FIVE MORE CONFIRMATIONS, AND THE CONSTANT-IN-`$v0` CASE | | L24570 | §224 | addendum (P31 S58b) — CROSS-JUMP MERGES *CALLS*, AND THE DELAY SLOT IS THE DISCRIMINATOR | | L24600 | §225 | addendum (P31 S58b) — THE GUARD-CLAUSE FINGERPRINT, AND THREE MORE SHAPES | | L24627 | §226 | addendum (P31 S58b) — THE FRAME CATALOGUE: SEVEN MORE LEVERS, AND SLOT ORDER IS DECLARATIO | | L24672 | §229 | addendum (P31 S58b) — NAME IT **INSIDE** THE ARM | | L24683 | §230 | CROSS-CONFIRMED (P31 S58b) | | L24692 | §231 | addendum (P31 S58b) — FOUR MORE WAYS THE LISTING MISLEADS | | L24723 | §232 | CROSS-CONFIRMED (P31 S58b) | | L24734 | §259 | THE DISCARD LEDGER FOR THE aa–bg HARVEST (P31 S58b): WHAT WAS MINED AND REJECTED, AND WHY | | L24773 | §260 | THE §154-A LEADING-ISLAND SPLIT: ONE CONFIG LINE, AND THE ISLAND PEELS FROM THE END (P31 S | | L24823 | §260-A | STAGE 2 IS PROVEN, AND THE WHOLE jtbl PIPELINE IS AUTOMATED AT THE GATE (P31 S59, same day | | L24859 | §261 | THE -O0 ORACLE: DERIVE THE OPT LEVEL FROM THE TARGET, AND `$fp` IS NOT THE TELL (P31 S59) | | L24885 | §261a | THE -O0 FRAME-RELOAD GRAMMAR: reload COUNT disambiguates the C spelling (P31 S59, byte-pro | | L24907 | §262 | A LANE'S YIELD IS ONLY A LANE FACT IF IT IS SIZE-MATCHED (P31 S59) | | L24937 | §263 | A STOLEN DELAY SLOT WHOSE INSTRUCTION IS AN ARG-REGISTER COPY IS AN ARITY ERROR, NOT A SCH | | L24978 | §264 | FOUR TELLS-LANE C RECIPES, EACH DRIVEN TO MATCH (P31 S59) | | L25028 | §265 | THE VERBATIM-ASM BANK LANE: A FUNCTION NO -O2 C CAN EVER MATCH BANKS AS A RAW `__asm__` BO | | L25090 | §266 | THE INERT-RIDER LAW: A LEVER IS ONLY CITABLE WHEN ITS SOLO REMOVAL BREAKS THE MATCH (P31 S | | L25130 | §267 | ADDENDA HARVESTED FROM WAVES at/bh/bk/bl (P31 S59b) | | L25138 | ADD-1 | → §231 addendum (also cross-ref from §195-D) — THE MASKED-`jal` "MISSING CALL" ILLUSION | | L25147 | ADD-2 | → §42a addendum — A SHARED CONSTANT *NAMED IN A LOCAL* ACROSS A `jal` IS AN ISO→TU DRIFT H | | L25159 | ADD-3 | → §236, item 10 — THE UN-DELETED `INCLUDE_ASM` STUB IS A DUPLICATE DEFINITION | | L25169 | ADD-4 | → §225-3/-4 addendum — THE MIRROR ROW: VALUE-RETURN IN THE *TAKEN* ARM + TRAILING BARE `re | | L25183 | ADD-5 | → §1/I1 addendum — THE INVERTED RANGE TEST: `(u32)(x-lo) >= N` WITH THE ZERO-ARM AS THE TR | | L25192 | ADD-6 | → §172b-1 / §264 addendum — SHIFT-AS-TEST: `(x << 16) != 0` TESTS THE LOW HALF WITHOUT TRU | | L25201 | ADD-7 | → §256 addendum — THE SINGLE-GUARD GOTO: THEN-BLOCK OUT OF LINE AT THE TAIL | | L25213 | ADD-8 | → §245 addendum — THE RE-TIE BARRIER PINS CALL-ARG SETUP TO SOURCE ORDER (construct 6) | | L25225 | ADD-9 | → §213-3 / §217 addendum — ADJACENT PRE-CALL FRAME STORES ARE ONE AGGREGATE; THE UNESCAPED | | L25235 | ADD-10 | → §237 addendum (arity-evidence paragraph) — AN `la` PAIR ABOVE THE PROLOGUE `sw $ra` IS A | | L25244 | ADD-11 | → Cross-confirmation card block (per §259's standing instruction: confirmation, not news) | | L25295 | §268 | A `register __asm__` PIN ON A CALL-CLOBBERED REGISTER IS HONORED EXACTLY WHEN THE PINNED R | | L25349 | §269 | ADDENDA HARVESTED FROM WAVES ax/bm (P31 S59c) | | L25364 | ADD-1 | → §257-8 addendum — THE INTERPOSED ASM'S `__volatile__` IS A PER-SHAPE DIAL, AND THE "NO-O | | L25377 | ADD-2 | → §236 addendum (item 1 corollary) — THE SILENT-SPLICE DECLARATION LADDER: WHERE THE TU HA | | L25391 | ADD-3 | → §238 addendum — TWO BINARIES CAN EACH DEFINE THE SAME `func_` NAME WITH DIFFERENT BYTES, | | L25404 | ADD-4 | → §167-37 addendum — THE FIFTH PRECONDITION, NOW MEASURED: DOWNSTREAM FIELD RE-READS ⇒ NAM | | L25421 | ADD-5 | → §267-ADD-6 / §172b-1 addendum — THE SIGN-TEST FACE: `(s16)v < 0` IS `sll 16` + `bgez` ON | | L25437 | ADD-6 | → §220-addendum — THE FOURTH FACE: WHEN THE NAMED COPY ITSELF BUYS THE EXTRA CALLEE-SAVED, | | L25453 | ADD-7 | → §229 addendum — THE VALUE FACE: A LOOP-STORE CONSTANT SPELLED AS A LITERAL HAS NO SOURCE | | L25466 | ADD-8 | → §30a addendum — THE THIRD COLUMN: `*p++` IS ALSO SERIALIZED BUT BURNS `addiu`; ONLY CAST | | L25481 | ADD-9 | → §255 "AND CASE-BODY PLACEMENT" bound / §222-addendum-3 — ON A LARGE SPARSE TREE, BODIES | | L25495 | ADD-10 | → Cross-confirmation card block (per §259's standing instruction: confirmation, not news) | | L25536 | §3-1a. | The §266 sweep — every solo-lever A/B run for this batch | | L25569 | §270 | The A-prop 0-bank anatomy: a byte-correct body still needs FOUR layers to agree (P31 S59) | | L25587 | §271 | Ordinal IMM pairing: text order is NOT emission order; emit CANDIDATES, let the oracle pic | | L25609 | §272 | The `(void)`-decl + empty-call wall: K&R the DEFINITION, not just the decls (P31 S59) | | L25629 | §273 | A standalone compile is the WRONG oracle for a TU-destined draft (P31 S59) | | L25644 | §274 | ADDENDA HARVESTED FROM 18 WAVES (P31 S60): 315 candidates, 255 already covered, 21 sharpen | | L25650 | ADDENDUM | to §179-C — the `.type NAME, @function` requirement | | L25697 | ADDENDUM | to §134 — a typedef defined BELOW the splice point is stripped anyway | | L25727 | ADDENDUM | to §179-D (GTE macro reference family) — `gte_SetRotMatrix`/`gte_SetTransMatrix` bodies | | L25753 | ADDENDUM | to §82 — struct copies must stay MEMBER-WISE, not block-moved, to match a spill-slot save | | L25791 | ADDENDUM | to §136d-1 (RC-12, the `$0`-add / opaque-copy family) — two symptoms beyond "compare reads | | L25830 | ADDENDUM | to §20 — a global declared as `T *` may itself BE the array base, not a pointer to derefer | | L25859 | ADDENDUM | to §1-I5 | | L25924 | ADDENDUM | to §164-51 | | L25940 | ADDENDUM | to §176-F5 | | L25969 | ADDENDUM | to §225 | | L26014 | ADDENDUM | to §224 — CROSS-JUMP: THE DUPLICATE CAN BE A PLAIN STORE, NOT ONLY A CALL | | L26048 | ADDENDUM | to §164-64 — AN EMPTY CLOBBER ON AN ARGUMENT REGISTER CAN BE THE DELIBERATE FIX, NOT JUST | | L26079 | ADDENDUM | to §244 — THE COMPILER BARRIER, NOT `volatile`, BLOCKS A STORE→LOAD HOIST THROUGH A POINTE | | L26110 | ADDENDUM | to §172b-1 — TWO PLACEMENT DIALS THE PROMOTION LAW DOESN'T NAME: NARROW THE COUNTER'S DECL | | L26157 | ADDENDUM | to §229 — A POINTER'S NUMBER OF USES DECIDES WHETHER ITS ADDRESS FOLDS OR SURVIVES A CALL | | L26202 | ADDENDUM | to §172b-4 — THE PLAIN CAST-DIVISION ALREADY PRODUCES THE PATTERN; DON'T HAND-ROLL THE BIA | | L26232 | ADDENDUM | to §8 — WHEN `INCLUDE_RODATA` NEEDS A STANDALONE `.s` YOU CAN'T CREATE, CARRY THE FRAGMENT | | L26275 | ADDENDUM | to §226 — A DEAD LOCAL SCOPED TO A NESTED BLOCK BUYS A MID-FUNCTION `addiu sp` PAIR, NOT A | | L26314 | ADDENDUM | to §172a — RE-READING MEMORY (NOT NAMING A TEMP) IS WHAT KEEPS AN INCREMENT'S DELAY-SLOT F | | L26345 | ADDENDUM | to §225 / §256 — A GOTO TO A SHARED SET-POINT PREVENTS IF-CONVERSION FROM COLLAPSING A LAT | | L26387 | ADDENDUM | to §211 — AN IN-LOOP ACCUMULATOR WANTS A CLOSED-FORM EXPRESSION WHEN THE TARGET REMATERIAL | | L26422 | §275 | THE LEFTOVER-REGISTER READ | | L26462 | §276 | MIXED ADDRESS-EXPRESSION SPELLING FOR ADJACENT RELOCATABLE SYMBOLS IS A CSE-UNIFICATION DI | | L26571 | §277 | RETURN-TAIL C SPELLING PICKS THE DELAY-SLOT-FILL vs TRAILING-MOVE TOPOLOGY, AND A NARROWER | | L26635 | §278 | ADDENDA HARVESTED FROM WAVE cf (P31 S60): 34 candidates, 13 already covered, 8 sharpenings | | L26643 | ADDENDUM | to §74 (func_800D0E30, resident) | | L26679 | ADDENDUM | to §172b-4 (func_80185054, ov_SC03_097) | | L26717 | ADDENDUM | to §265 (func_8017DC80, ov_SC07_002) | | L26756 | ADDENDUM | to §17 (func_800CB794, md_MAIN_036) | | L26804 | ADDENDUM | to §162p (func_8017C120, ov_MAIN_012) | | L26838 | ADDENDUM | to §20 (~L1947) (func_80186AD0, ov_SC06_032) | | L26868 | ADDENDUM | to §164-56 (func_8017F644, ov_SC04_005) | | L26903 | ADDENDUM | to §237 (func_8017F7FC, ov_SC03_092) | | L26932 | §279 | A `do/while (p < end)` LOOP UNDER AN ENTRY GUARD: the guard decides the compare, not the l | | L26933 | §NNN | A DO-WHILE'S GUARD AND LATCH MUST REPEAT THE SAME BOUND EXPRESSION, NOT SHARE ONE LOCAL: T | | L26966 | §280 | THE CURSOR'S DECLARED TYPE PICKS `sltu` vs `slt`: a `T*` bound test is UNSIGNED by C rule, | | L26967 | §NNN | A LOOP CURSOR'S C TYPE (POINTER vs PLAIN INTEGER) SELECTS `sltu` vs `slt` FOR ITS BOUND TE | | L27000 | §281 | GROUP COPY-THEN-RMW BY OPERATION KIND, NOT FIELD BY FIELD: sched1 does the interleaving, t | | L27001 | §NNN | A DEPENDENT COPY-THEN-RMW BLOCK MUST BE SOURCE-GROUPED BY OPERATION KIND, NOT BY FIELD, AN | | L27046 | §282 | gcc's OWN LOOP REVERSAL PUTS THE COUNTER INIT AFTER THE HOISTED MOVABLES — a position no h | | L27047 | §NNN | WRITE THE UP-COUNT LOOP: gcc's OWN REVERSAL PRODUCES A COUNTER-INIT INSTRUCTION THAT LANDS | | L27084 | What | I could not verify | | L27131 | §283 | ADDENDA HARVESTED FROM THE 36-WAVE BATCH (P31 S60): 1,216 candidates, 859 already covered, | | L27141 | ADDENDUM | to §164-75 — the fold-reassociation law also fires at a variable's INITIALIZER, not only a | | L27167 | ADDENDUM | to §21 — the `bltz`+`slti` (or N-separate-compares) signed-range-split bullet is now CONFI | | L27213 | ADDENDUM | to §202 — the DEF-SIDE ALIAS also resolves a function-vs-DATA-symbol identifier clash, not | | L27236 | ADDENDUM | to §22 (`volatile`-qualified-global reload lever, cookbook ~L1922) — for a NON-constant, s | | L27255 | ADDENDUM | to §195-E — a goto-ladder's STORES must sit AT the labels, after the gotos, not inline bef | | L27317 | ADDENDUM | to §195-N — a GNU statement-expression slider must sit INSIDE the conditional arm's value | | L27374 | ADDENDUM | to §176-B2 — in a micro-function with no long/short lifetime asymmetry, BOTH contending ps | | L27421 | ADDENDUM | to §199-G — A `default:` LABEL GROUPED ONTO THE LAST CASE REMOVES THE `j default` TAIL, EV | | L27490 | ADDENDUM | to §220 — REFERENCING THE RAW PARAMETER (NO NAMED COPY, NOT EVEN A PIN) LETS THE CALLEE-SA | | L27548 | ADDENDUM | to §252 — a `>=0`/`<0` split on an unconditionally-decremented value needs the POSTFIX ope | | L27575 | ADDENDUM | to §215 — FIFTH SHAPE: reused mask constants across two call-free merge sites each get the | | L27658 | What | I could not verify | | L27729 | Harness-defect | flags | | L27822 | ADDENDUM | to §167-40 (func_8017E2CC, ov_SC04_015 — wave dg) | | L27842 | ADDENDUM | to §20's cross-jump EXPLOIT bullet (func_8017E360, ov_SC05_007 — wave dg) | | L27856 | ADDENDUM | to §87 (func_801815F4 ov_SC06_032; corroborating func_801840DC ov_SC05_017, func_80189C68 | | L27875 | ADDENDUM | to §153 / §236-5 (func_801A419C, md_SC07_003 — waves di and dl, corroborating; refutes a c | | L27895 | ADDENDUM | to §263 (func_801E83AC, md_SC04_029 — wave dj) | | L27909 | ADDENDUM | to §265 (func_8017D878, ov_SC03_107 — wave dj; corroborated by a REJECTED, contradicted ca | | L27923 | ADDENDUM | to §6 (func_801811F0, ov_SC03_102 — waves dj and dl, corroborated by a self-reported "noth | | L27937 | ADDENDUM | to §195-G (func_80183BB0, ov_SC05_001 — wave dj) | | L27965 | ADDENDUM | to the zero-byte-asm-slider family (§47 / §148-C / §153) (func_800CB874, md_MAIN_040 — wav | | L27983 | ADDENDUM | to §194-B (func_8017EB34, ov_SC03_117 — wave dk; distinct from the §74 co-pinning finding | | L28015 | ADDENDUM | to §74 (func_8017EB34, ov_SC03_117 — counter/clamp variant; wave dj, corroborated by dk/dl | | L28035 | ADDENDUM | §37 — merged into the §153/§236-5 entry above (func_801A419C, wave dl) | | L28043 | ADDENDUM | §6 — merged into the §6 entry above (func_801811F0, wave dl) | | L28051 | ADDENDUM | to §238 (func_80182CB4, ov_SC02_000 — wave dl) | | L28069 | ADDENDUM | to §137a (func_801684B4, ov_MAIN_012; corroborated independently by func_80189E68, func_80 | | L28085 | ADDENDUM | to §8c / §88d (func_8016AB6C, ov_MAIN_012 — wave dm) | | L28099 | ADDENDUM | to §73 / §30#2 (func_800D1984, resident — wave dm) | | L28115 | What | I could not verify | | L28135 | ADDENDUM | to §174 Law 4 (func_8017E9A8, ov_SC06_015) | | L28175 | ADDENDUM | the zero-emission-asm family gains a REF-SLIDER PLACEMENT LAW and a paired RESTORER (func_ | | L28227 | From | ck + cl + cm | | L28233 | ADDENDUM | §NNN — sharpens §55a / §164-37 / §165-28 (switch-vs-tree cluster): A SHAPE THAT LOOKS LIKE | | L28270 | ADDENDUM | §NNN — sharpens §167-13's boundary: CHAINING TWO IDENTICAL SIDE-BY-SIDE STORES INTO ONE C | | L28303 | ADDENDUM | §NNN — sharpens §162a3/§60b: A `sll $v0,16 / sltiu $v0,1` ZERO-TEST OF A JUST-DECREMENTED | | L28336 | ADDENDUM | §NNN — sharpens §37/§124 (unspecified-parameter-list family): DERIVE A CALLEE'S ARITY LOWE | | L28369 | ADDENDUM | to §5a (func_80181F74, ov_SC03_112, wave cn) | | L28419 | ADDENDUM | to §265 (func_8017E26C, ov_SC04_016, wave cn) | | L28457 | ADDENDUM | to §42b (func_8018247C, ov_SC07_002, waves cu + cw) | | L28488 | ADDENDUM | to §199-F family (func_8017EFB0, ov_SC02_021, wave cu) | | L28543 | ADDENDUM | to §195-E (func_800CFC1C, md_MAIN_003, wave cv) | | L28588 | ADDENDUM | to §215 addendum (func_800CB2C8, md_MAIN_033, waves cv + cw) | | L28644 | ADDENDUM | to §236 item 4 (func_8017D268, ov_SC04_006, wave cw) | | L28673 | What | I could not verify | | L28705 | Harness-defect | flags (not idioms — flagged for the operator) | | L28746 | ADDENDUM | to §250 (func_8017D7CC, ov_SC03_115 — cx/cy/cz/dr) | | L28791 | ADDENDUM | to §225 (func_8017E190, ov_SC03_115 — cx/cy/cz) | | L28844 | ADDENDUM | to §45-A (func_8017F6A4, ov_SC02_016 — cy/cz) | | L28888 | ADDENDUM | to §249 (func_80182ED4, ov_SC04_004 — dp/dr/dt) | | L28933 | ADDENDUM | to §199-A (func_801816FC, ov_SC02_005 — dp/dr/dt) | | L28979 | What | I could not verify | | L29022 | Harness-defect | flags | | L29084 | §284 | COMBINE CAN REASSOCIATE TWO SEQUENTIAL BITWISE-AND MASKS INTO ONE AGAINST THE PRE-MASK VAL | | L29088 | §285 | PRE-INITIALIZING A VARIABLE WITH A SHARED CONSTANT BEFORE A BRANCH MAKES BOTH ARMS OF THE | | L29092 | §286 | FOLD A STATEMENT'S SIDE EFFECT INTO A COMMA-EXPRESSION IN AN ARGUMENT POSITION TO PLACE IT | | L29096 | §287 | A STACK-FRAME HOLE BELOW TWO ADDRESS-TAKEN AGGREGATE LOCALS IS A LEADING PAD MEMBER OF ONE | | L29100 | §288 | A REGISTER PIN DECLARED UNINITIALIZED AND ASSIGNED ONLY AT ITS LATE, SOLE USE STILL FORCES | | L29104 | §289 | an array local's address-taken base keeps every element's store alive, even though only on | | L29108 | §290 | A SINGLE STRENGTH-REDUCED GIV CAN DRIVE STORES TO SEVERAL DISTINCT RELOCATABLE SYMBOLS, EA | | L29112 | §291 | THE DELAY-SLOT FALSE-VALUE: A CONDITIONAL BRANCH'S ZERO ARM MUST BE A FALL-THROUGH-ADJACEN | | L29116 | §292 | DECLARING A SYMBOL UPSTREAM OF AN ALREADY-BANKED SIBLING THAT RELIES ON THAT SYMBOL'S IMPL | | L29119 | §293 | THE "WHOLE-OBJECT GATE NEEDS EVERY SIBLING" LAW, WRITTEN PROPERLY: IT IS THREE LAWS, AND T | | L29169 | §294 | ADDENDA HARVESTED FROM THE S61 GEN0/MAXTOK/MAIN/DEEPSEEK BATCH (waves ab8/ab16/ab24/ab32 · | | L29180 | ADDENDUM | to §127 — the -O0 constant-offset fold keys on the MEMBER-ACCESS tree shape, not on the in | | L29208 | ADDENDUM | to §261a — FOUR MORE -O0 DIALS BEYOND THE RELOAD COUNT (boot.c + md_MAIN_011/003, wave m0a | | L29258 | ADDENDUM | to §264-3 — the explicit entry copy's BOUND: required exactly when the RAW value must outl | | L29278 | ADDENDUM | to §172b-1 (counter-type dial) — the sll's SOURCE REGISTER is the placement discriminator | | L29290 | ADDENDUM | to §276 — the SHARE direction: spell the second adjacent symbol RELATIVE to force ONE anch | | L29306 | ADDENDUM | to §31's density-dummy dial (L2460/L2497) — the dose is TWO refs, and the dummy must sit w | | L29323 | ADDENDUM | to §282 — the ADDRESS-GIV face: write the walked address INLINE so strength reduction birt | | L29346 | ADDENDUM | to §253 — SECOND byte-proven card (upgrade from single-observation), and the placement fac | | L29367 | ADDENDUM | to the L1800 anchor-steer bullet — a DERIVED-POINTER local silently flips the merged giv's | | L29390 | ADDENDUM | to §31's asm→layout inference — TWO WIDTH-PAIR DISCRIMINATORS (same offset, different widt | | L29411 | §295 | THE KERNEL-TRAP STUB: §81's jr-DETECTOR WITHOUT A TABLE IS A PsyQ SYSCALL TRAMPOLINE — ROU | | L29466 | §296 | THE FRAME CHECK OUTRANKS THE ATLAS LEVER: READ PROLOGUE/EPILOGUE BEFORE DRAFTING ANY C — A | | L29509 | §297 | ONE GIV SERVING MIXED-WIDTH LOADS *AND* A STORE NEEDS STRUCT-MEMBER SPELLING; CAST/INDEX S | | L29552 | §298 | THE SHARED-TAIL POSITION DIAGNOSTIC: A FOLDED TAIL *BETWEEN* SWITCH ARMS MEANS PER-ARM DUP | | L29577 | §299 | TWO INDEPENDENT EXTRACTION CHAINS EMIT CONTIGUOUSLY INSIDE ONE EXPRESSION; ONLY A STATEMEN | | L29612 | §300 | S61 DISTILL BATCH NOTES (waves ab8/ab16/ab24/ab32 · g0a–g0f · m0a · ds1/ds2 · m0b) | | L29681 | §301 | AN INTERNAL `j` CARRIES `R_MIPS_26 .text`: rtu/match_one "MATCH" COULD NOT SEE WHICH LABEL | | L29729 | §302 | A RED BINARY IS A DRIFTED SPEC, NOT A MYSTERY: THE THREE CARVE-STATE INVARIANTS AND HOW TO | | L29773 | §303 | MODULE ISLAND TABLES: DERIVE THE PADS AT BUILD TIME, PEEL NOTHING — THE §154-A/§260 "islan | | L29809 | §304 | SELF-DEFINING RODATA: WHEN A FUNCTION'S `.s` IS THE ONLY OWNER OF THE DATA IT REFERENCES, | | L29832 | §305 | "CARVE-REFUSED" AT GATE TIME IS THREE NAMED, DETERMINISTIC CLASSES — NONE OF THEM A CARVE | | L29884 | §306 | A HAZARD `nop` IN FRONT OF A DIV-RESULT STORE IS A STATEMENT-ORDER DEFECT: THE INDEPENDENT | | L29911 | §306a | T4 DISTILL ADDENDA (P31 S62; four byte-proven refinements to existing laws, verified again | | L29949 | §307 | THE BRUTE-FORCE-THE-STATEMENT-ORDERS LEVER HAS A BOUND: A FAN-OUT COPY'S PRIORITY IS SCHED | | L29993 | §308 | A NAIVE FINAL `goto` DELETES THE TARGET'S `j`+`nop`: SPELL THAT EDGE AS A TAUTOLOGICAL `if | | L30022 | §308a | A TWO-INSTRUCTION ARM STUB LAID *BEFORE* THE OTHER ARM AND REACHED ONLY BY A TRAILING `j` | | L30056 | §309 | A FRAME-ADDRESS EQUIVALENCE RIDES *THROUGH* A SKIPPABLE CONDITIONAL: POST-`if` STORES FOLD | | L30096 | §310 | A TWO-OPERAND `subu`'s DESTINATION TIES ONLY TO THE OPERAND THAT IS BLOCK-LOCAL: SHARE *ON | | L30127 | §311 | A JOIN-BLOCK STORE CAN SWALLOW AN ARM'S LOAD-DELAY `nop`: WHEN ONE ARM'S VALUE COMES FROM | | L30169 | §312 | A BARE RELATIONAL IN AN `if` GIVES THE `slt` NO TARGET: ASSIGN THE COMPARISON INTO THE ALR | | L30195 | §313 | A RECOVERY STAGE THAT REWRITES THE DRAFT IS PART OF THE MEASUREMENT: GATE THE RAW DRAFT FI | | L30251 | §314 | AN IF/ELSE-IF ABS-RANGE CHECK'S C SPELLING IS INERT: FOUR STRUCTURALLY DIFFERENT REWRITES | | L30283 | §315 | ALL-CONSTANT AGGREGATE FILL: THE EMISSION ORDER IS SHARED-LITERAL GROUPS × DESCENDING INDE | | L30320 | §316 | A GUARD CHAIN THAT GATES A SUCCESS BLOCK ROUTES ITS FAILING GUARDS TO A LABEL **BEFORE** T | | L30354 | §317 | A NARROW STRUCT-FIELD STORE TURNS A *LIVE SImode VARIABLE* INTO A `li`: convert.c's TRUNCA | | L30388 | §318 | A UNARY MINUS STORED BACK INTO THE SAME HALFWORD IS COMPUTED IN HImode, SO ITS LOAD IS `lh | | L30463 | Addendum | §148-A2 addendum — the empty-asm insn_count dial (N bare `__asm__("")` = exactly +N real i | | L30471 | Addendum | §312 addendum — naming the compare's dest is the vehicle, the hard-reg pin on that name is | | L30475 | Addendum | subu/sh dest reuses a pinned operand's dying register (func_8017F498) | | L30483 | Addendum | Addendum to §261a — the -O0 increment-operator copy also fires on the POINTER LOCAL itself | | L30487 | Addendum | Addendum to §164-XX/§165-06 — at -O0 a bare local IS a memory lvalue, so the increment ope | | L30491 | Addendum | Addendum to §312 — the compare's named destination must itself carry a hard register: nami | | L30495 | Addendum | Addendum to §167-12 — the single-operand volatile keepalive, re-confirmed at a gated MATCH | | L30499 | Addendum | A §5a BARRIER SITS IN REORG'S FILL WINDOW TOO, SO ITS SIDE OF THE INNER `goto` IS CHOSEN P | | L30503 | Addendum | Loop sentinel wrongly hoisted out of a call-containing loop (func_801835B0) | | L30515 | Addendum | Reversed-operand sltu then xori 1 means unsigned less-or-equal (func_8017D948) | | L30519 | Addendum | Chained *2*2 array index folds to one copy;sll, not two (func_80011380) | | L30533 | Addendum | REGALLOC-PERM: one memory-clobber barrier insufficient — stack two (func_80182D1C) | | L30539 | Addendum | Last if-arm's nested if un-fuses the whole preceding chain (func_8017FE88) | | L30543 | Addendum | A plain global-RMW triplet is a fillable scheduling window too — zero drift, and the fille | | L30547 | Addendum | Shared arithmetic that CONSUMES the selected value cannot be hoisted above the `if` — narr | | L30551 | REFUTED | claim — Dead call-arg register address-fold: no C-level lever found (func_8017ECAC) | | L30570 | Addendum | Truncating assign must be the lazy `||` operand, not hoisted (func_80012B58) | | L30582 | Addendum | WIDTH sh!=lhu residual: hoist a sibling field's load first (func_80182538) | | L30599 | REFUTED | claim — REFUTED — the `j`-slot store is an ASPSX macro-expansion hop, not a cse split; mas | | L30601 | §NNN | REFUTED: A `sw %lo(SYM)($at)` IN A JUMP'S DELAY SLOT WITH ITS `lui $at` STRANDED ABOVE THE | | L30625 | Addendum | Static local_type blocker survives typedef removal — it's textual (func_801588CC) | | L30632 | Addendum | Orphan sh triplet to $sp is a write-only local array (func_8017F274) | | L30636 | Addendum | REGALLOC-PERM: the store reads the narrow copy's register — split the one narrow local by | | L30640 | Addendum | REGALLOC-PERM store in a cross-jump shared tail defeats §137's barrier (func_8017EDE8) | | L30644 | Addendum | LENGTH-DRIFT nop clears when offset math precedes the symbol launder (func_80039B20) | | L30648 | REFUTED | claim — REFUTED: a block shared across TWO switch statements is ordinary forward cross_jum | | L30650 | §NNN | REFUTED: A CODE BLOCK SHARED BY TWO *DIFFERENT* SWITCH STATEMENTS IS ORDINARY FORWARD `cro | | L30666 | RETRIEVAL | FAILURES this round (not new knowledge — a RETRIEVAL defect) | | L30681 | §319 | N TEXTUALLY DUPLICATED `return v;` TAILS PUT THE LAST `or` AND THE RETURN-REGISTER MOVE IN | | L30710 | Addendum | Masked-OR field-merge plateaus at close=10; bitfield store clears it (func_8017FD64) | | L30714 | Addendum | Integer-space address arithmetic is a THIRD no-movable spelling, and a distant `SYM[0]` re | | L30724 | Addendum | LENGTH-DRIFT −1 on a coalesced-away copy: a CALLER-SAVED SOURCE pin can resurrect it, boun | | L30731 | Addendum | Return-0 statement order: extra j+move vs delay-slot fusion (func_8018BB50) | | L30735 | Addendum | Counter zero in the DECLARATION slot, empty for-init — the prologue SHIFT-DRIFT/+K face of | | L30753 | Addendum | The dying-pinned-register reuse on a sign-extend chain: route every later read through a s | | L30757 | Addendum | Volatile zero-byte asm slider scrambles the prologue schedule (func_801814B0) | | L30767 | Addendum | Register-pinned helper pointer local regresses closeness; use plain local (func_80013CFC) | | L30771 | Addendum | §194-A addendum — the bare/colon-less fence measured NULL and only the "memory"-clobber fo | | L30775 | RETRIEVAL | FAILURES this round | | L30796 | §320 | THE §43 "RETURN-TYPE FLIP PAIR" IS **NOT** TU-EDIT-REQUIRED: THREE DRAFT-ONLY ESCAPES (P31 | | L30855 | §321 | FILE-SCOPE DUPLICATE ANONYMOUS-STRUCT TYPEDEFS ARE A HARD ERROR; THE SAME TEXT AT BLOCK SC | | L30876 | §322 | A PROBE THAT ANSWERS A *NECESSARY BUT NOT SUFFICIENT* QUESTION WILL PRICE BLOCKED WORK AS | | L30906 | §323 | CARRYING FILE-SCOPE TYPES BETWEEN SPLIT TUs: DEDUPE BY NAME, SKIP HEADER-PROVIDED, AND THE | | L30936 | §323a | R53, TWICE IN ONE HOUR: A FAILED BUILD LEAVES THE PREVIOUS BINARY, AND `sha1sum` READS GRE | | L30945 | §323b | A SCRIPT THAT PARSES argv AT IMPORT CANNOT BE SHARED; EXTRACT THE PREDICATE, DO NOT COPY I | | L30959 | §324 | THE ARITY-SELF WALL NEEDS *BOTH* HALVES: A K&R DEFINITION **AND** A NO-PROTO TU DECL; EITH | | L30997 | §325 | REGALLOC: A SHARED SMALL CONSTANT STORED TWICE IN THE PRE-LOOP BLOCK IS A *local-alloc* $s | | L31010 | §326 | DEFEATING ADDRESS-CSE: SPELL TWO READS OF THE SAME HALFWORD DIFFERENTLY AND GCC CANNOT SHA | | L31019 | §327 | A RANGE TEST MUST BE HImode: WITH `s32` + A `(u16)` CAST GCC *PROVES* THE MASK REDUNDANT A | | L31034 | §328 | THE VOLATILE ALIAS MUST BE AN *OBJECT*, NOT A CAST: `*(volatile s32*)&sym` UNFOLDS %lo INT | | L31044 | §329 | fold-const NARROWS `(int)s16_var & 0xFFF` ONTO THE *RAW HImode PSEUDO*, BREAKING THE SIGN- | | L31052 | §330 | THE NEIGHBOUR-SHAPE LEVER IS THE CHEAPEST FIRST MOVE, AND IT IS UNDER-USED (P31 S67; four | | L31069 | §331 | OPEN GAP: NO LEVER ELIMINATES AN *UNWANTED DUPLICATE* COPY AT A BRANCH-TARGET BLOCK HEAD ( | | L31081 | §332 | THE maspsx `la`-IN-A-DELAY-SLOT GAP: gcc EMITS A SYMBOLIC ADDRESS LOAD AS ONE ATOMIC lengt | | L31106 | §332a | MAIN'S RESIDUAL FRONTIER IS CONTAMINATED WITH TOOLCHAIN WALLS: 4 OF 7 IN ONE WAVE (P31 S67 | | L31125 | §333 | FRAME SIZE IS SET BY *DECLARED* AGGREGATES, NOT USED ONES: AN UNREFERENCED TRAILING LOCAL | | L31139 | §334 | A RELOAD SPILL SLOT IS ROUNDED TO `BIGGEST_ALIGNMENT` (8B), SO ONE SPILLED 4-BYTE PSEUDO C | | L31150 | §335 | AN `extern u16 A[]` READ AT A VARIABLE SUBSCRIPT ALLOCATES DEAD STACK TEMPS (~8B PER ACCES | | L31159 | §336 | THE §5a CROSS-JUMP BARRIER GOES AT THE *BOTTOM* OF THE TWIN, NOT THE TOP: `find_cross_jump | | L31167 | §337 | THE CC1-ONLY BLOCKER CLASS: `blocker_probe`'s STATIC ORACLE REPORTS "none" AND THE WHOLE-T | | L31181 | §338 | `jtbl_carve._sltiu_bounds` MISREADS A NON-SWITCH `sltiu` AS A BOUNDS CHECK, OVER-SPANNING | | L31191 | §339 | A 2-CASE SWITCH OMITS THE LOW-BOUND RANGE TEST, SO THE PRESENCE OF `slti/bnez` BETWEEN THE | | L31219 | §340 | §194-K COROLLARY: **FLIP THE FALSE EDGE YOU CANNOT DELETE.** A "scheduler" residual can be | | L31253 | §341 | AN HImode STORE TEMP REWEIGHTS A sched2 TIE-BREAK THAT NO STATEMENT ORDER CAN REACH (P31 S | | L31267 | §342 | A `void *` PARAMETER CAST TO ITS REAL TYPE IN A LOCAL IS **NOT** BYTE-NEUTRAL WHEN A LATER | | L31284 | §343 | `decl_prior`'s FLEET MAJORITY CAN BE WRONG ABOUT THE TRUE SIGNATURE — READ THE RIVALS, NOT | | L31306 | §344 | RAISE A BIV'S global_alloc PRIORITY WITH A ZERO-BYTE REFERENCE INSTEAD OF PINNING IT; PINN | | L31322 | §345 | A VOLATILE **STORE** EVICTS THE MEM FROM cse AND KEEPS `sh`; A VOLATILE **LOAD** BLOCKS co | | L31332 | §346 | `c ? X : -X` TAKES expand_expr's COND_EXPR **SINGLETON** PATH (copy, then negate IN PLACE) | | L31352 | §347 | LOOP REGISTER ASSIGNMENT IS A **DECLARATION-ORDER + LIVE-RANGE** DIAL: FIVE COMPOSABLE LEV | | L31382 | §347-addendum | A THIRD INSTANCE, AND THE SHARPEST STATEMENT OF THE RULE (md_MAIN_025/func_800CB300, 243 i | | L31400 | §343-addendum | SECOND INSTANCE OF THE WRONG-MAJORITY DECL (same function) | | L31407 | §348 | THE BASE SPELLING PICKS THE ADDRESSING MODE: A SYMBOL GIVES THE 3-INSN `lui/%lo` FORM, A P | | L31424 | §349 | RE-ASSIGN A BASE POINTER AT THE END OF THE LOOP BODY TO MAKE `n_times_set > 1` — THAT KILL | | L31445 | §350 | A ZERO-BYTE RE-TIE SETS `reg_n_sets=2`, WHICH KILLS sched1's `birthing_insn_p` LAUNCH_PRIO | | L31466 | §351 | `/s` (MEM_IN_STRUCT_P) IS A DIAL YOU CHOOSE PER ACCESS: A COMPONENT_REF GRANTS IT AND LETS | | L31496 | §352 | ⚠ TWO IDENTICAL `__asm__ __volatile__("")` BARRIERS **MERGE WITH EACH OTHER** — SPELL THE | | L31529 | §353 | USE `-fno-thread-jumps` AS AN **ORACLE** TO PROVE A RESIDUAL IS thread_jumps, THEN LAUNDER | | L31552 | §354 | THE giv **WORTH-WHILE TEST** IS A DIAL: RE-ASSOCIATE THE ADDEND INTO THE INDEX TERM AND `s | | L31570 | §355 | A REMAPPED SIBLING'S **SOURCE BIAS IS NOT ITS EMITTED BIAS** — DO NOT HAND-SHIFT OFFSETS T | | L31584 | §356 | MEASURE A DRAFT IN THE TU IT WILL LIVE IN, NOT IN THE STANDALONE PROBE (P31 S68; measured | | L31599 | §357 | ONE STRUCT POINTER, NOT TWO: A SECOND SOURCE VARIABLE BUILDS A THIRD IV (P31 S68; byte-pro | | L31608 | §358 | (sharpens §333) — AN **UNREFERENCED** FIXED-SIZE AGGREGATE LOCAL IS LOAD-BEARING (P31 S68; | | L31618 | §359 | (§43-adjacent) — SPELL A SIGN-WIDEN AS AN EXPLICIT TWO-STEP, FUNCTION-SCOPED TEMP (P31 S68 | | L31631 | §360 | THE "COMPILER FOUND A SHORTER EQUIVALENT THAN THE TARGET" PAIR (P31 S68; main/func_800241C | | L31647 | §361 | ★ — A LOOP-TAIL BYTE SIGNATURE THAT NAMES ITS SOURCE SHAPE — AND THE LAW THAT A "SCHEDULIN | | L31692 | §362 | TWO TRAPS WHEN A CARVE MOVES A STUB INTO THE `-O0` TU (P31 S68; byte-proven, 6 fns / 2,547 | | L31715 | §363 | ★★ — THE OVERLAY-LAYOUT ASSUMPTION IS A SYSTEMIC BUG CLASS, AND `main` IS THE EXCEPTION TH | | L31744 | §364 | ★ — THE libgpu `P_TAG` BITFIELD SPELLING IS **OPT-LEVEL DEPENDENT** (P31 S68; two function | | L31758 | §365 | PIN **BOTH** MASKS OR NEITHER (P31 S68; ov_SC01_000/func_8017E594, 357 ins) | | L31767 | §366 | ★★ — `group_case_nodes` MERGES **STACKED CONSECUTIVE** CASE LABELS: GIVE EVERY CASE ITS OW | | L31787 | §367 | RECONCILING A DECL CONFLICT BETWEEN TWO DRAFTS FOR THE SAME TU (P31 S68; main) | | L31802 | §368 | ★★★ — THE **RELOAD-REMAT CONSTANT**: REACH A REGISTER NO PIN CAN REACH (P31 S68; ov_SC03_1 | | L31832 | §369 | REUSE THE **COMPARE CONSTANT'S OWN VARIABLE** FOR A MASK THAT KEEPS COALESCING (P31 S68; m | | L31848 | §370 | ★★ — A **HARD BOUND** FROM sched.c, AND THE reorg SLOT-STEAL DIAGNOSTIC (P31 S68; main/fun | | L31900 | §371 | ★★ — CARVING A **SINGLE-OBJECT MODULE BINARY**, AND THE spimdisasm RODATA-MIGRATION TRAP T | | L31949 | §372 | ★★★ — THE **COPY-CAPTURE PAIR**, AND THE ONE ZERO-BYTE EDIT THAT DEFEATS BOTH (P31 S68; by | | L31987 | §373 | ★★★ — THE **DEAD-RESET CSE-BREAKER**, THE PIN THAT BREAKS A sched2 ANTI-DEP, AND WHY AN `a | | L31989 | §3-1. | DEAD-RESET CSE-BREAKER — the zero-footprint replacement for a §195-I asm re-tie | | L32007 | §3-2. | A REGISTER PIN THAT DELETES A sched2 ANTI-DEPENDENCE | | L32020 | §3-3. | HARD FACT FOR THE SCHEDULING MAP — an `asm` ALWAYS has priority 1 | | L32028 | CROSS-REFERENCE | TO §370 — checked and found INAPPLICABLE here, which is the point | | L32036 | §374 | A `register … __asm__("$30")` RESERVATION IS **NOT HONOURED** BY `move_movables` (P31 S68; | | L32052 | §375 | AN `$a0`-`$a3` PIN USED LATE RELOCATES AN **EARLIER** OUTGOING-CALL USE OF THAT REGISTER ( | | L32065 | §376 | ★★★ — A STANDALONE `match_one` CLOSENESS OF 0 IS A CLAIM ABOUT THE **BODY**, NEVER ABOUT T | | L32107 | §377 | THREE HARNESS DEFECTS FOUND IN ONE GATING SESSION, ALL "A CONFIDENT NUMBER ABOUT A SMALLER | | L32124 | §378 | ★★★ — THE **SELF-CALLER CAST**: LET A TU KEEP CALLING THE FUNCTION IT IS ABOUT TO DEFINE ( | | L32174 | §379 | ★★★ — **MEM_IN_STRUCT_P**: THE SAME LOAD, WRITTEN AS A STRUCT MEMBER, SCHEDULES WHERE A CA | | L32205 | §380 | ★★★ — **A SECOND SET OF A PSEUDO DISQUALIFIES IT FROM `move_movables`** (P31 S69; main/fun | | L32226 | §381 | THE `insn_count` HOIST THRESHOLD IS A DIAL YOU CAN READ WITH `cc1 -dL` (P31 S69; four inde | | L32242 | §382 | TWO FOLD REASSOCIATIONS THAT NEED THEIR OWN STATEMENT (P31 S69) | | L32254 | §383 | TWO TOOLCHAIN FACTS THE PACKS DID NOT CARRY (P31 S69) | | L32269 | §384 | ★★★ — A CARVE-CONFIG BANK IS RED UNTIL YOU RE-EXTRACT, AND THAT LOOKS EXACTLY LIKE A FALSE | | L32319 | §385 | ★★★ — THE **SCHED2 PRIORITY-DONOR ASM**: closing the "hoisted-invariant vs IV-init prehead | | L32345 | §386 | ★★★ — A BYTE LOAD ON THE **BIV** BASE WAS BORN IN THE COMBINE PASS: SPELL IT AS A SHIFT-MA | | L32371 | §387 | ★★ — **SPLIT-FOLD DISPATCH CLOBBER**: one switch case needs a reload, another must keep th | | L32394 | §388 | ★★★ — THE **-O0 COLOURING ORACLE**: simulate `stupid.c` instead of grinding spellings (P31 | | L32419 | §389 | ★★★ — `h_norm` IS BLIND TO INDEXED-GLOBAL RELOCS, SO FREE WORK BECOMES AN INVISIBLE SINGLE | | L32460 | §390 | ★★★ — MINIMUM DISTANCE IS NOT MINIMUM WORK; RANK TWIN CANDIDATES BY EFFORT, AND FILTER LOO | | L32498 | §391 | ★★ — A BYTE-ALIGNED STRUCT COPIES IN FOUR INSTRUCTIONS, A WORD-ALIGNED ONE IN TWO (P31 S69 | | L32512 | §392 | ★★★ — THE SONNET-WAVE HARVEST (P31 S69): SEVEN TYPE/ORDER LEVERS THAT EACH CLOSED A MATCH | | L32553 | §393 | ★★ — THE **BIRTHING BOOST**: A SINGLE-SET LOCAL IS SCHEDULED LAST; GIVE IT A SECOND SET (P | | L32570 | §394 | ★★ — TWO ALIGN-1 ACCESSES IN ONE FUNCTION RESERVE A PHANTOM STACK SLOT (P31 S69; ov_SC02_0 | | L32582 | §378b | ★★★ — THE FOUR VARIANTS OF THE DECL BLOCKER, AND THE TWO PLACES §378 DOES **NOT** APPLY (P | | L32621 | §395 | ★★★ — FIVE NARROWING/PLACEMENT LEVERS FROM ONE 91-INSTRUCTION CRACK (P31 S69; byte-proven | | L32655 | §314b | ★★ — TWO ABS-RANGE GUARDS IN ONE FUNCTION NEED **DIFFERENT SPELLINGS** (P31 S69; byte-prov | | L32677 | §322b | ★★★ — THE CARVE CLASS IS COMPLETABLE, AND EVERY WORKTREE `CARVE-REFUSED` WAS AN INSTRUMENT | | L32712 | §332b | ★★★ — THE §332 "WALLS" ARE A PER-OBJECT ASSEMBLER MODE, NOT A C LIMIT — 6 CLOSE AS REAL C | | L32732 | §378c | ★★ — THE FIFTH DECL-BLOCKER VARIANT: THE DRAFT REDECLARES WHAT THE TU ALREADY OWNS (P31 S6 | | L32747 | §396 | ★★★ — SIX LEVERS FROM THE S69 SINGLETON ROUND, INCLUDING ONE THAT ONLY A `COND_EXPR` REACH | | L32795 | §397 | ★★★ — RE-RUN THE TWIN SCAN AFTER EVERY EXEMPLAR BANK; A CLUSTER SIBLING IS FREE THE MOMENT | | L32828 | §396g | ★★ — A GUARD LADDER'S RUNGS MUST STAY SYMMETRIC OR `reorg.c` LOSES ITS BRANCH REDIRECT (P3 | | L32845 | §398 | ★★★ — `family_remap` CARRIES THE **SOURCE** TU's DECL ENVIRONMENT INTO A DESTINATION THAT | | L32875 | §398b | ★★ — NAMING A SUB-EXPRESSION IN A LOCAL CHANGES WHICH PSEUDO SURVIVES; INLINE IT AT BOTH U | | L32902 | §399 | ★★★ — FOUR LEVERS FROM THE FINAL S69 ROUND (P31 S69; each byte-proven, none previously in | | L32942 | §400 | ★★ — A BASELINE CHECK THAT CONFLATES "ABSENT EVERYWHERE" WITH "CHANGED UNDER US" SILENTLY | | L32977 | §401 | §401 | | L33010 | §402 | §402 | | L33034 | §403 | §403 | | L33066 | §404 | §404 | | L33102 | §405 | ★★★ — THE S70 WAVE HARVEST: 130 agents, 113 MATCH, and the laws they brought back | | L33108 | §3-A. | THE ORACLE HAS A HOLE: match_one cannot see a jump table (§405-A) | | L33118 | §3-B. | THE SCHEDULER DIALS (the dominant residual family this wave) | | L33134 | §3-C. | REGISTER ALLOCATION FROM C, WITHOUT PINS | | L33150 | §3-D. | INTEGRATION IS STILL THE BOTTLENECK, AND THE TU IS THE AUTHORITY | | L33158 | §3-E. | WHAT THE AGENTS REFUTED | | L33168 | §406 | ★★ — THE PROLOGUE-WEAVE CLASS IS MECHANICAL AND LARGE (134 open stubs) | | L33190 | §407 | ★★ — LATE-WAVE ADDENDA TO §405 (the last agents in) | | L33219 | §408 | ★★★ — §406 REFUTED AS A SWEEP: THE SHAPE IS THE FAMILY, THE DISAGREEMENT IS THE DEFECT (P3 | | L33271 | §409 | ★★★ — THE S71 JOURNAL-FUELLED WAVE: 100% FIRST-PASS MATCH, AND THE NINE LAWS IT BROUGHT BA | | L33298 | §3-The | nine laws this wave produced | | L33363 | §410 | ★★★ — COPY THEN ACCUMULATE ON THE COPY: resolving the birthing-boost vs register-allocatio | | L33396 | §411 | ★★★ — THE PACK MUST CARRY THAT FUNCTION'S OWN HISTORY (P31 S71; measured 38/39 vs 124/131) | | L33434 | §412 | ★★★ — §323 CARVE BLOCKER 2 WAS A REGEX THAT COULD NOT SEE PAST `__attribute__` (P31 S71) | | L33483 | §413 | ★★★ — DIFFICULTY IS THE RESIDUAL CLASS, NOT `nins` — ROUTE THE MODEL TIER OFF HISTORY (P31 | | L33520 | §414 | ★★★ — `parallel_gate` ON `main` IS A FALSE PASS, AND THE RULE WAS ALREADY WRITTEN DOWN (P3 | | L33557 | §415 | ★★ — A FILE-SCOPE DECL MAKES gcc-2.7.2 MERGE THE TU'S LATER *BLOCK-SCOPE* EXTERNS INTO IT | | L33582 | §416 | ★★ — FOUR LEVERS FROM THE S71 OVERNIGHT LANE, none of which the cookbook held (P31 S71) | | L33615 | §417 | ★★★ — A REGISTER PIN CAN BLOCK `jump.c`'s SELECT COLLAPSE, AND UNPINNING THEN EXPOSES A `c | | L33642 | §418 | ★★★ — TWO LOOP-STRUCTURE LEVERS: MAKE THE SECOND INDEX A GIV, AND KEEP A TABLE ADDRESS UNF | | L33668 | §419 | ★★★ — WHEN A PIN IS IMPOSSIBLE, WIN THE local-alloc DENSITY CONTEST INSTEAD (P31 S71; byte | | L33702 | §420 | ★★★ — A MULTI-CLUSTER SYMBOL REBASE, AND THE BARE-NAME DEDUP THAT HID THREE QUARTERS OF IT | | L33739 | §421 | ★★★ — A `la $tN` + `addiu` PAIR CAN BE A **RELOAD** ARTIFACT THAT NO C SPELLING REACHES (P | | L33770 | §422 | ★★ — QImode ARITHMETIC VIA `(u8)(x - K)`, AND `flag ^ 1` NEEDS ITS OWN TEMP (P31 S71; byte | | L33790 | §423 | ★★★ — "MATCH IN ISOLATION + GATE REJECTS + CAUSE NOT DETERMINED" ⇒ GREP THE TU FOR A FILE- | | L33817 | §424 | ★★★ — EQUAL-PRIORITY STORES COME OUT **REVERSED**: sched1's LUID tie picks the LAST statem | | L33839 | §425 | ★★★ — `sb` ALIASES SCALAR GLOBALS WHILE `sh`/`sw` STRUCT STORES DO NOT, AND TWO MORE ALIAS | | L33865 | §426 | ★★★ — main's SWITCH FUNCTIONS WERE NEVER A CODEGEN WALL: ONE RODATA CARVE HAD BEEN MISSING | | L33945 | §427 | ★★ — A HASH IS A CORRECTNESS ORACLE WITH ZERO DIAGNOSTIC CONTENT; PRESERVE THE RED ARTIFAC | | L33964 | §428 | ★★★ — A ZERO-BYTE CROSS-JUMP BARRIER: ADVANCE THE POINTER INSIDE EACH SWITCH ARM (P31 S72; | | L34002 | §428a | ★★★ — TWO RESIDUALS THAT MOVE IN OPPOSITE DIRECTIONS UNDER EVERY LEVER USUALLY SHARE ONE C | | L34050 | §429 | ★★★ — EVERY HELD POINTER NEEDS ITS OWN LOCAL, AND A NEGATIVE-DISPLACEMENT BYTE STORE NEEDS | | L34083 | §430 | ★★★ — A GOTO INTO A LOOP IS FINE; HAND-HOIST THE CONSTANTS IT COSTS YOU (P31 S73 — **this | | L34132 | §431 | ★★★ — SPLITTING A 27,000-LINE TU AT ITS ORIGINAL BOUNDARIES: THE JTBL SPANS TELL YOU WHERE | | L34195 | §433 | ★★★ — ON A SWITCH FUNCTION, CASE SOURCE ORDER IS THE DOMINANT RESIDUAL — AND `match_one` I | | L34227 | §432 | ★★★ — DEFEAT cse's MERGE OF TWO IDENTICAL MASKS BY SPELLING ONE AS A SHIFT PAIR (P31 S72/S | | L34269 | §434 | ★★★ — TWO SYMBOLS, ONE FRAME: RUN THE FRAME CHECK BEFORE DRAFTING ANYTHING LARGE (P31 S73; | | L34317 | §435 | ★★★ — AN OVERLAY TU SPLIT IS NEAR-FREE, AND THE GAP TEST IS "IS THIS WORD A CODE ADDRESS", | | L34374 | §436 | ★★★ — TWO TOOLS THAT READ THE WRONG SOURCE OF TRUTH, AND THE SHAPE THEY SHARE (P31 S74; bo | | L34431 | §437 | ★★★ — A WRAPPED TRAILING COMMENT BROKE THE TU PARSER, AND THE §8b DECL LAYER WROTE `extern | | L34538 | §438 | ★★★ — THE SAME-ADDRESS LEAD IS NOW SIZE-FILTERED: A HOMONYM IS WORSE THAN NO TWIN (P31 S74 | | L34565 | §439 | ★★ — LEVER SET FROM THE S74 WAVE (each entry is one measured crack, not a hypothesis) | | L34586 | §440 | ★★★ — A `.rodata` CARVE PIECE BINDS TO A SUBSEG, NOT TO A FUNCTION: EXTEND THE CARVE INSTE | | L34627 | §441 | ★★ — THREE MORE INSTRUMENT DEFECTS FROM THE SAME SESSION, ALL OF WHICH BLAME THE SUBJECT ( | | L34657 | §442 | ★★★ — A RECOVERY RUNG THAT REWRITES A LEGAL CONSTRUCT INTO AN ILLEGAL ONE READS EXACTLY LI | | L34712 | §443 | ★★ — A DERIVED DEPENDENCY MUST BE PROVISIONED BY *EVERY* PROVISIONER, AND WE HAVE TWO (P31 | | L34739 | §444 | ★★★ — A REJECTED GATE DAMAGED COMMITTED CARVE STATE, AND `git status src/` SAID NOTHING WA | | L34788 | §445 | ★★ — `make clean BINARY=` IS FLEET-WIDE: THE VARIABLE IS ACCEPTED AND IGNORED (P31 S75) | | L34817 | §446 | ★★★ — "RELOC-ONLY REMAP GATED DIFF" IS A CARVE VERDICT, NOT A CODEGEN VERDICT, UNTIL YOU D | | L34864 | §447 | ★★★ — THE BIGGEST FUNCTION LEFT IS A CARVE, AND THE VERDICT TOOL SAID "DECLARATIONS" (P31 | | L34908 | §448 | ★★★ — ASSEMBLY POSING AS C: 154 GAME FUNCTIONS THE REPORTS COUNTED AS DONE (P31 S75) | | L34951 | §449 | ★★★ — FOUR COMPILER DIALS FROM THE S75 REDRAFT WAVE (two byte-gate CONFIRMED, and one corr | | L34956 | §3-A. | `reg_n_sets` IS A ONE-LINE SCHEDULING DIAL — `t = t + 1;` vs `*p = t + 1` (CONFIRMED: `ov_ | | L34985 | §3-B. | A SINGLE-SET LOCAL'S VALUE IS VISIBLE AT A SWITCH JOIN, AND THAT ERASES A ZERO-EXTENSION ( | | L35000 | §3-C. | CORRECTION TO §439 — `sll 16; srl 16` LANDS *AFTER* THE CALL, AND WORKS FOR A KNOWN CONSTA | | L35018 | §3-D. | AN OFFLINE JTBL-RODATA PLACEMENT AUDIT (CLAIMED: `md_MAIN_034:func_800CB00C` — did NOT ban | | L35031 | §450 | ★★★ — REGENERATING A TARGET `.s` FOR A FUNCTION THAT IS NO LONGER A STUB (P31 S75; 146 fun | | L35072 | §451 | ★★★ — YOUR EVIDENCE HAS MORE THAN ONE SOURCE, AND THE ONE YOU QUERY IS PROBABLY THE WORSE | | L35110 | §452 | ★★★ — NOT EVERY VERBATIM BODY IS UNDECOMPILED WORK, AND §448'S HEADLINE OVERSTATED IT (P31 | | L35147 | ADDENDUM | to §265 (P31 S76) — THE PROSE GUARD DID NOT HOLD. IT IS A GATE REFUSAL NOW. | | L35192 | CORRECTION | to §182/§188 (P31 S76) — THE EPILOGUE "WALL" IS AN ORACLE ARTIFACT IN THE REORDER ISLAND | | L35232 | §460 | `-dS` PRINTS THE SCHEDULER'S READY LIST WITH PRIORITIES. STOP INFERRING IT FROM RTL ORDER. | | L35264 | §461 | LAUNDERING AN INVARIANT CAN BE THE DEFECT, AND "RESIDUAL A" DOES NOT GENERALIZE PAST ONE B | | L35294 | §465 | THE ASPSX SLOT-HOP: A GAP OUR REORDER SUBSTITUTE CANNOT CLOSE (`main:func_8005F830`) | | L35320 | §466 | `main` (509 ins, -O0): ADDRESS CONTEXT EMITS `mult` INDEX-FIRST | | L35338 | §464 | FOUR VOLATILE/BARRIER LEVERS FROM `main:func_8005DE78` (141 ins → MATCH) | | L35366 | §467 | GLOBAL-ALLOC TIES BREAK ON *DECLARATION* ORDER, AND A COPIED CLOBBER LIST IS A DEFECT | | L35390 | §468 | THE `%lo`-FOLD EXTENDS TO STORES ONLY VIA `extern Struct SYM[]`, AND MASKING HID THE OPERA | | L35424 | §469 | THE `MEM_IN_STRUCT_P` ALIAS UNLOCK (and §463's spill law, independently confirmed) | | L35447 | §470 | FOUR CSE/SCHED LEVERS FROM `main:func_800301C8` (170 ins, 133 → 18) | | L35475 | §471 | A LAUNDER'S REAL COST IS AN ALLOCNO, AND `$t0` IS RELOAD'S (`main:func_80032A74`, 408 → 12 | | L35499 | §472 | 🔴 §148-A's HOIST THRESHOLD IS 29, NOT 58, WHEN THE LOOP CONTAINS A CALL | | L35529 | §473 | 🔴 §265's "HANDWRITTEN" VERDICT FOR `ov_SC07_002:func_8017DC80` IS REFUTED (324 → 89) | | L35554 | §474 | A *PROVED* C-LEVEL FLOOR: `split_tree` + `stupid.c` (-O0), from `main:func_80011380` | | L35591 | §475 | THE `"memory"` FENCE AS A *CSE INVALIDATOR*, AND `(b*3)<<3` INSTEAD OF `b*24` | | L35622 | §476 | 🔴 A HARD-REGISTER PIN DESTROYS TWO THINGS COMBINE AND SCHED1 NEED (`func_800226C0`, 670 in | | L35650 | §462 | FOUR LEVERS FROM `main:func_80024054` (91 ins, 74/53/32 → 4) | | L35687 | §463 | 🔴 SPILL SLOTS ARE 8 BYTES, AND THE §41b "LOAD ABOVE THE PROLOGUE" WALL IS REFUTED | | L35723 | §477 | ★★★ — THE `self_decl_tu` CLASS IS A SOLVED, MECHANICAL LANE: 16 DRAFTS, 16 BANKS (P31 S77) | | L35797 | §478 | 🔴 — A VERBATIM DRAFT IS THE STRONGEST *FALSE* SIGNAL YOUR SCOPING TOOL CAN EMIT | | L35817 | §479 | ★★★ — WHERE THE PERMUTER ACTUALLY PAYS: A MEASURED YIELD CURVE (P31 S77, 8 candidates) | | L35896 | §480 | 🔴 — A STATIC BLOCKER CLASS THAT THE REAL PIPELINE ALREADY REMOVES IS A PHANTOM | | L35911 | §481 | ★★★ — `conflicting types` IS A SAME-SCOPE ERROR; ACROSS SCOPES IT IS ONLY A WARNING (P31 S | | L35942 | §482 | ★★★ — TWO INDEPENDENT RE-TIES, ORDERED: WHEN ONE BARRIER FIXES ONE RESIDUAL AND CREATES TH | | L35977 | §483 | ★★★ — THE S77w WAVE HARVEST: SIX LEVERS, FOUR FROM BANKED (BYTE-PROVEN) BODIES | | L36033 | §484 | ★★★ — "NO SINGLE NOLOAD BASE" IS NOT "UNLINKABLE": ASK WHETHER THE `.bss` OFFSETS ARE DISJ | | L36079 | §485 | ★★★ — THE PLACEMENT MAP WAS PARSING A PRETTY-PRINTER: 25 PsyQ OBJECTS WERE INVISIBLE, NOT | | L36115 | §486 | ★★★ — CARVING AN `-O0` ISLAND IN **main**: FIVE COUPLED PIECES, AND THE TWO THAT ANNOUNCE | | L36160 | §487 | ★★★ — THE PSX LOADER'S PER-VERSION SIGNATURE SETS ARE A FREE PROVENANCE ORACLE: main's "WA | | L36212 | §488 | ★★ — THE "GAME CODE" GAPS BETWEEN LIBRARY BLOCKS WERE LIBRARY OBJECTS: 13 SUBSEGS → LINKED | | L36252 | §489 | ★★★ — SCATTERED `.bss` IS A PARTITION PROBLEM: SPLIT THE SECTION INTO PER-BASE PIECES AT L | | L36310 | §490 | ★★★ — THE "WALL" BAND WAS A LIBRARY VERSION AWAY: LIBPAD 4.2.1 + LIBAPI 4.2 FOUND, THE WHO | | L36356 | §491 | ★★ — THE MECHANICAL LEFTOVERS (P31 S79 #6): A PHANTOM STUB, TWO JTBL TWINS, ONE EXACT CLON | | L36402 | §492 | ★★ — "C-PLUMBING" WAS THREE DIFFERENT THINGS (P31 S79 #7): A RAW SPLICE THE GATE'S LADDER | | L36430 | §493 | ★★ — THE PERMUTER ROUTE END-TO-END, AND THE THREE PLUMBING STEPS BETWEEN A SCORE-0 WINNER | | L36490 | §494 | ★★★ — TEN BANKS FROM ONE-AGENT-PER-FUNCTION DRAFTING (P31 S79/S80 #9): THE IDIOMS, THE PLU | | L36630 | §495 | ★★★ — THE VERBATIM END-STATE (P31 S80 #10): TWO DEF-SIDE DECLARATION WALLS, A "BANK" THAT | | L36671 | §496 | ★★ — A CARRIED-TYPE TEST THAT ASSUMES THE OVERLAY INCLUDE SET SILENTLY DROPS A RESIDENT TY | | L36705 | §497 | ★ — A BODILESS `typedef struct Tag Alias;` DEFINES THE ALIAS, NOT THE TAG: THE CARRIER'S F | | L36728 | §498 | ★★★ — A CARVE THAT DROPS THE BINARY'S `--pre` CLAUSE, AND A GATE THAT IGNORED THE EXTRACT' | | L36758 | §499 | ★★★ — A NEVER-ONBOARDED PAYLOAD PINS ITS OWN BASE STATICALLY, AND THE FIRST BUILD CANNOT ( | | L36799 | §500 | ★★★ — THE T3 WAVE HARVEST (P32 T3, 2026-09-05): 31 one-agent-per-function drafters → 20 MA | | L37157 | §501 | ★★★ — A LEVER THAT MEASURES WORSE MAY BE A CASCADE: READ THE `.loop` DUMP FOR THE DESIRABI | | L37561 | §453 | ★★★ — ONE SOURCE PER UNIQUE FUNCTION: THE INCLUDE-AT-SITE SHARE, ITS GATE, AND THE FOUR DE | | L37610 | §454 | Rung D and rung R: taking a lever off a body that still has to compile to the same bytes ( | | L37677 | §454a | the recipe rung's real shape: replication, not discovery (S99, measured) | | L37692 | §455 | Rung G, the guided search: score the object, classify the residual, compose the moves (Pha | | L37765 | §456 | The lever-removal move catalog: what the S104 agents found under ~205 pins, barriers and l | | L37814 | §457 | The lever-removal move catalog, S105: eight TU batches, 42 of 42 at 0 (Phase 36 T7) | | L37886 | §458 | Struct spelling is a per-access dial: the T2 probe's byte evidence (Phase 37 S106) |