diff --git a/src/ov_SC07_006/ov_SC07_006.c b/src/ov_SC07_006/ov_SC07_006.c
index 1574d4382..c30a1a5f8 100644
--- a/src/ov_SC07_006/ov_SC07_006.c
+++ b/src/ov_SC07_006/ov_SC07_006.c
@@ -922,7 +922,58 @@ DEFINE_func_8012B70C()  /* dedup: shared engine-core @0x8012b70c (src/shared) */
 DEFINE_func_8012B744()  /* dedup: shared engine-core @0x8012b744 (src/shared) */
 
 
-INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006", func_8012B77C);
+// @class: struct
+// @stuck: none — MATCH (58 ins)
+#include "common.h"
+
+extern s32 ratan2(s32 a0, s32 a1);
+extern s32 func_80047948(s32 a0); /* rsin-like: angle (0..0xFFF) -> 1.12 fixed */
+extern s32 func_8004787C(s32 a0); /* rcos-like: angle (0..0xFFF) -> 1.12 fixed */
+
+/* The 4-byte destination is a PACKED PAIR of angles held entirely in ONE saved register
+ * ($s4) across all three calls — that is what produces the target's
+ *   andi $s4,$s4,0xFFFF   (read of the still-uninitialized local, hoisted into the prologue)
+ *   ... or  $s4,$s4,ang<<16      -> t.hi = yaw
+ *   ... and $s4,$s4,0xFFFF0000 / or $s4,$s4,ang&0xFFFF  -> t.lo = pitch
+ * A `struct { s16 lo, hi; }` has align 2 -> BLKmode-ish handling: gcc spills it to the stack
+ * and stores it with lwl/lwr + swl/swr (54 ins, 53 mismatched). Two 16-bit BITFIELDS in a
+ * u32 container give the SImode, align-4 struct gcc keeps in a register. */
+typedef struct {
+    u32 lo : 16;
+    u32 hi : 16;
+} Ang2_8012B77C;
+
+/* a0 = destination packed-angle word, a1 = "from" entity, a2 = "to" entity.
+ * Both entities carry 16.16 fixed-point x/y/z at +0/+4/+8; the s16 reads at +2/+6/+0xA are
+ * the integer halves. Returns the destination pointer — the return value is REAL: without it
+ * gcc stores with `sw $s4,0($s3)` (57 ins) instead of the target's
+ *   addu $v0,$s3,$zero ; sw $s4,0($v0)   (the return-value copy that the store's base coalesces onto).
+ * The flattened `dz` reuse at the end is also load-bearing: making the (dz*sin + dx*cos)>>12
+ * temp its OWN variable gives it $a1 for the whole chain; reusing `dz` extends that allocno so
+ * it lands in $s5 exactly as the target does (`sra $s5,$v0,12` / `negu $a1,$s5`), and it also
+ * demotes dz's priority so the saved-reg order comes out $s3=out, $s4=t, $s5=dz. */
+s32 func_8012B77C(s32 out, s32 from, s32 to) {
+    Ang2_8012B77C t;
+    s32 dx, dy, dz;
+    s32 ang;
+    s32 r1, r2;
+
+    dz = *(s16 *)(to + 0xA) - *(s16 *)(from + 0xA);
+    dx = *(s16 *)(to + 0x2) - *(s16 *)(from + 0x2);
+    dy = *(s16 *)(to + 0x6) - *(s16 *)(from + 0x6);
+
+    ang = (ratan2(-dz, dx) - 0x400) & 0xFFF;
+    t.hi = ang;
+
+    r1 = func_80047948(ang);
+    r2 = func_8004787C(ang);
+    dz = (dz * r1 + dx * r2) >> 12;
+    t.lo = ratan2(dy, -dz);
+
+    *(Ang2_8012B77C *)out = t;
+    return out;
+}
+
 
 DEFINE_func_8012B864()  /* dedup: shared engine-core @0x8012b864 (src/shared) */
 
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_80131340.c b/src/ov_SC07_006/ov_SC07_006_jr_80131340.c
index 648945c86..33ffb9102 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_80131340.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_80131340.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_80135888.c b/src/ov_SC07_006/ov_SC07_006_jr_80135888.c
index a92446efe..2e70e3e5d 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_80135888.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_80135888.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_80135A4C.c b/src/ov_SC07_006/ov_SC07_006_jr_80135A4C.c
index edc4d4dc3..07fb5bcd3 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_80135A4C.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_80135A4C.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_80135D20.c b/src/ov_SC07_006/ov_SC07_006_jr_80135D20.c
index 8037fc62a..708b4e4de 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_80135D20.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_80135D20.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
@@ -1006,7 +1006,80 @@ void func_80136EC4(void) {
 DEFINE_func_80136ECC()  /* dedup: shared engine-core @0x80136ecc (src/shared) */
 
 
-INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006_jr_80135D20", func_80136F3C);
+// @class: regalloc-order
+// @stuck: none — MATCH (match_one 61/61, relocation-masked)
+//
+// 2-line "X cross" GsSortLine family (same shape as byte-proven DEFINE_func_80137030 /
+// DEFINE_func_80137178, RED variant):
+//   line1 = (a0-3, a1-3) -> (a0+3, a1+3);  line2 = (a0+3, a1-3) -> (a0-3, a1+3);
+//   color r=0xFF g=0 b=0.  All four endpoints are loop-invariant across the two calls, so all
+//   six live values (x0,y0,x1,y1,white,base) are homed in $s0-$s5 for the whole body.
+//
+// Levers that make gcc-2.7.2 -O2 emit the target's exact regalloc + schedule:
+//   * ROLE COALESCING (the 28 -> 13 step): the target coalesces x0 into a0's incoming copy
+//     ($s1: `addu $s1,$a0,$zero` ... `addiu $s1,$s1,-0x3`) and y1 into a1's copy ($s2), while
+//     x1 lands in a FRESH $s0. gcc coalesces whichever derived value is computed LAST off the
+//     param, so the source must emit y0, x1, y1, x0 in that order — writing x0 first (the
+//     "natural" order) coalesces x1 instead and costs 28 instructions.
+//   * `__asm__("" : "=r"(x1) : "0"(x1))` right after `x1 = a0 + 3;` nails x1's compute point
+//     between x1 and y1, which is what keeps x1 from being the last a0-user (13 -> 7).
+//   * PINS: x1 -> $s0 ($16), x0 -> $s1 ($17), base(&D_800A6518) -> $s3 ($19), white(0xFF) ->
+//     $s4 ($20). Without the x1 pin the whole $s bank rotates by one; without the x0 pin the
+//     a0-copy fights x1 for $s0. y0/y1 must stay UNPINNED — pinning y1 to $s2 kills the
+//     `addu $s2,$a1,$zero` param copy entirely (a1 stays in $a1, -1 insn, 58 mismatched).
+//   * NO anchor on `white`: an `__asm__` on white front-loads its `li $s4,255` ahead of the
+//     call's $a0/$a2 argument setup (the last 4-mismatch schedule residual). Leaving white
+//     un-anchored lets the scheduler emit `addiu $a0,$sp,0x10` / `addu $a2,$zero,$zero` first,
+//     exactly as the target does.
+extern short D_800B9A02;
+extern u8 D_800A6518[];
+extern void GsSortLine(void *a0, void *a1, s32 a2);
+
+void func_80136F3C(s32 a0, s32 a1)
+{
+    struct {
+        u32 tag;
+        s16 x0;
+        s16 y0;
+        s16 x1;
+        s16 y1;
+        u8 r;
+        u8 g;
+        u8 b;
+    } line;
+    register s16 x1    __asm__("$16");               /* $s0 */
+    register s16 x0    __asm__("$17");               /* $s1 */
+    register u8 *base  __asm__("$19") = D_800A6518;  /* $s3 */
+    register s16 white __asm__("$20");               /* $s4 */
+    s16 y0;
+    s16 y1;
+
+    y0 = a1 - 3;
+    x1 = a0 + 3;
+    __asm__("" : "=r"(x1) : "0"(x1));
+    y1 = a1 + 3;
+    white = 0xFF;
+    x0 = a0 - 3;
+    line.tag = 0;
+    line.r = white;
+    line.g = 0;
+    line.b = 0;
+    line.x0 = x0;
+    line.y0 = y0;
+    line.x1 = x1;
+    line.y1 = y1;
+    GsSortLine(&line, &base[(u16)D_800B9A02 * 20], 0);
+    line.tag = 0;
+    line.r = white;
+    line.g = 0;
+    line.b = 0;
+    line.x0 = x1;
+    line.y0 = y0;
+    line.x1 = x0;
+    line.y1 = y1;
+    GsSortLine(&line, &base[(u16)D_800B9A02 * 20], 0);
+}
+
 
 DEFINE_func_80137030()  /* dedup: shared engine-core @0x80137030 (src/shared) */
 
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_8013F350.c b/src/ov_SC07_006/ov_SC07_006_jr_8013F350.c
index f25e0c68c..dd3af0a10 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_8013F350.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_8013F350.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_80140608.c b/src/ov_SC07_006/ov_SC07_006_jr_80140608.c
index b57b1598a..06bf90dc6 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_80140608.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_80140608.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
@@ -2372,7 +2372,100 @@ void func_80146360(void)
 }
 
 
-INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006_jr_80140608", func_801463A0);
+// @class: struct
+// @stuck: none — MATCH (101 ins); also byte-verified spliced into the real ov_SC07_006 TU
+/* func_801463A0 — ov_SC07_006 / ov_SC07_006_jr_80140608 (cross-overlay family reach 138).
+ *
+ * Byte-verified twice: `match_one` in isolation AND spliced into a scratch copy of
+ * src/ov_SC07_006/ov_SC07_006_jr_80140608.c compiled whole (§42a: iso-MATCH != TU bank).
+ *
+ * Keys:
+ *  - D_80126B78 is `s32 *`, read ONCE before the five calls -> it lives in $s0 across them;
+ *    &D_80126B58 is CSE'd into $s1 for all eight `addu $a0,$s1,$zero` call arguments.
+ *  - The 8-byte BE0->BE8 copy is a STRUCT ASSIGN of an align-1 blob (cookbook §48-C2) ->
+ *    emit_block_move force_regs both bases (`la $a1` / `la $a0`) + lwl/lwr + swl/swr.
+ *    NOT `memcpy(...)`: the real TU declares `extern void *memcpy(...)`, which disables the
+ *    builtin TU-wide and would lower this copy to a `jal memcpy` CALL (§42a T1 drift class).
+ *  - NEW IDIOM (this was the whole residual, -1 instruction): the halfword store at the SAME
+ *    address as the block-copy SOURCE must not be spelled with the same C declaration. If it is,
+ *    cse substitutes the block move's already-loaded base register for the store's constant
+ *    address (`sh $v1,0($a3)` instead of `lui $at,%hi; sh %lo($at)`) — one instruction short, and
+ *    the extended live range of that base also pushes the block move off $a1/$a0 onto $a3/$a1.
+ *    Splitting the two views into two DECLS (here: the TU's `u8 D_80126BE0[]` for the copy, an
+ *    `__asm__("D_80126BE0")`-labelled `u16` for the store) gives cse two non-identical SYMBOL_REFs
+ *    (rtx_equal_p compares XSTR pointers), so the direct `lui %hi / sh %lo($at)` macro survives.
+ *    The emitted relocation is still against D_80126BE0 — no new undefined symbol.
+ *  - The three halfword stores must precede the D_80126B9C read-modify-write so its `sw` schedules last.
+ *  - Callee decls mirror the real TU's environment (the engine_core DEFINE_func_* signatures and
+ *    `u8 D_80126BE0[]` from DEFINE_func_80146128) so the def splices in with no conflicting types.
+ */
+#include "common.h"
+
+typedef struct { char _b[8]; } Blk8_801463A0;   /* size 8, align 1 -> unaligned block move */
+
+extern void func_8014C6F4(u8*);
+extern void func_80155150(s32 a0);
+extern void func_801470C0(s32 a0);
+extern void func_80147478(s32 a0);
+extern void func_80147118(s32 a0);
+extern void func_8014BDE8(s32 a0);
+extern short func_801508F8(s32 a0);
+extern void func_8014B5B0(s32 *a0);
+extern void func_80161D88(s32 a0);
+
+extern s32 D_80126B58;
+extern s32 *D_80126B78;
+extern u16 D_80126B5E;
+extern u16 D_80126B62;
+extern u16 D_80126B66;
+extern u16 D_80126C90;
+extern u16 D_80126C92;
+extern u16 D_80126C94;
+extern s32 D_80126B9C;
+extern s32 D_80126BA0;
+extern u8 D_80126BE0[];     /* identical to the decl DEFINE_func_80146128() already emits in this TU */
+extern u8 D_80126BE8[];
+extern u16 D_80126BE0_hw __asm__("D_80126BE0");   /* halfword view of BE0 — see NEW IDIOM above */
+extern u16 D_80126BE2;
+extern u16 D_80126BE4;
+extern s16 D_80126C9E;
+
+void func_801463A0(void)
+{
+    s32 obj;
+    short v;
+
+    obj = (s32)D_80126B78;
+    func_8014C6F4((s32)&D_80126B58);
+    func_80155150((s32)&D_80126B58);
+    func_801470C0((s32)&D_80126B58);
+    func_80147478((s32)&D_80126B58);
+    func_80147118((s32)&D_80126B58);
+    if (obj != 0) {
+        v = D_80126B5E + D_80126C90;
+        *(short *)(obj + 8) = v;
+        *(s32 *)(obj + 0x48) = (s32)v;
+        v = D_80126B62 + D_80126C92;
+        *(short *)(obj + 0xA) = v;
+        *(s32 *)(obj + 0x4C) = (s32)v;
+        v = D_80126B66 + D_80126C94;
+        *(short *)(obj + 0xC) = v;
+        *(u16 *)(obj + 0x2C) = *(u16 *)(obj + 0x2C) | 0x11;
+        *(s32 *)(obj + 0x50) = (s32)v;
+    }
+    func_8014BDE8((s32)&D_80126B58);
+    *(Blk8_801463A0 *)&D_80126BE8 = *(Blk8_801463A0 *)&D_80126BE0;
+    D_80126BE0_hw = D_80126B5E;
+    D_80126BE2 = D_80126B62;
+    D_80126BE4 = D_80126B66;
+    D_80126BA0 = D_80126B9C;
+    D_80126B9C = D_80126B9C & 0x3FFFFFFF;
+    func_801508F8((s32)&D_80126B58);
+    D_80126C9E = 0;
+    func_8014B5B0(&D_80126B58);
+    func_80161D88((s32)&D_80126B58);
+}
+
 
 DEFINE_func_80146534()  /* dedup: shared engine-core @0x80146534 (src/shared) */
 
@@ -3472,7 +3565,87 @@ DEFINE_func_8014C6D0()  /* dedup: shared engine-core @0x8014c6d0 (src/shared) */
 DEFINE_func_8014C6E0()  /* dedup: shared engine-core @0x8014c6e0 (src/shared) */
 
 
-INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006_jr_80140608", func_8014C6F4);
+// @class: schedule
+// @stuck: none — MATCH (match_one 91/91)
+//
+// Levers used (all byte-gated on ov_SC07_006):
+//  1. §struct  8-byte alignment-1 struct copy `*(M8_8014C6F4*)(a+0x164) = *(M8_8014C6F4*)(p+0x10)`
+//     -> the lwl/lwr,lwl/lwr,swl/swr,swl/swr block (engine_types.h M8_xxx convention).
+//  2. §17 base-pointer cache: `u8 *p = D_80078E78;` (NOT a direct D_80078E78[0x49]) so gcc pins
+//     the base into callee-saved $s1 and hoists the la into the prologue (live across the jal).
+//  3. §17 register pins: the search result `e` MUST be $v0 and the limit `lim` $v1, else
+//     local-alloc swaps the loop IV/limit pair ($v1<->$a0) and coalesces `e` into the IV.
+//  4. DELAY-SLOT lever (new): writing the compare constant as its own pre-loop statement
+//     (`want = 0x22;`) instead of the literal `0x22` inside the loop test. With the literal,
+//     loop.c hoists the `li $a1,0x22` into the loop PREHEADER (after the duplicated entry test),
+//     so reorg.c's guard branch has nothing local to take and instead COPIES the branch-target
+//     insn (`addu $v0,$zero,$zero`) into its delay slot and redirects past it — a +1 shift that
+//     also lets the loop-back branch steal the same insn (nop -> move). Materialising the
+//     constant in the block BEFORE the entry test makes fill_simple_delay_slots take it from
+//     the preceding insns, reproducing `beqz $v0,.L8014C7F4 / addiu $a1,$zero,0x22` and leaving
+//     the loop-back delay slot a nop. `want` needs no pin — gcc lands it in $a1 on its own.
+
+typedef struct { char _b[8]; } M8_8014C6F4;
+
+void func_8014C6F4(u8 *a) {
+    extern u8 D_80078E78[];
+    extern s32 D_8011F9D0;
+    extern s32 func_8016F1AC(void);
+    extern void func_80015978(s32 a0, s32 *a1);
+
+    u8 *p = D_80078E78;
+    register u8 *e __asm__("$2");
+    register u8 *lim __asm__("$3");
+    u8 *q;
+    s32 want;
+    u8 *src;
+
+    if ((*(u32 *)(a + 0x44) & 2) != 0) {
+        a[0x1C3] = 1;
+    }
+    if (((*(u32 *)(a + 0x44) ^ *(u32 *)(a + 0x48)) & *(u32 *)(a + 0x44) & 0x200) != 0) {
+        a[0x1C3] = 0;
+    }
+    *(M8_8014C6F4 *)(a + 0x164) = *(M8_8014C6F4 *)(*(u8 **)(a + 0x20) + 0x10);
+    if (func_8016F1AC() != 0
+        || (((*(u32 *)(a + 0x44) & 0x200) != 0) && a[0x1C3] == 0)) {
+        *(s16 *)(a + 0x162) = 1;
+        goto tail;
+    }
+    if (p[0x49] == 0x17) {
+        want = 0x22;
+        q = (u8 *)&D_8011F9D0;
+        lim = q + 0xC30;
+        for (; q < lim; q += 0x68) {
+            if (*(u16 *)q == want) {
+                e = q;
+                goto found;
+            }
+        }
+        e = 0;
+    found:
+        if (e != 0) {
+            src = e + 4;
+            goto call;
+        }
+    }
+    if (*(u16 *)a == 0x19) {
+        goto done;
+    }
+    src = a + 4;
+call:
+    func_80015978((s32)src, (s32 *)(a + 0x15C));
+done:
+    *(s16 *)(a + 0x162) = 0;
+tail:
+    if (*(s16 *)(a + 0x15A) == 0) {
+        *(u16 *)(a + 0x154) = *(u16 *)(a + 0x6);
+        *(u16 *)(a + 0x156) = *(u16 *)(a + 0xA);
+        *(u16 *)(a + 0x158) = *(u16 *)(a + 0xE);
+    }
+    *(s16 *)(a + 0x15A) = 0;
+}
+
 
 DEFINE_func_8014C860()  /* dedup: shared engine-core @0x8014c860 (src/shared) */
 
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_80154C24.c b/src/ov_SC07_006/ov_SC07_006_jr_80154C24.c
index b0be3d3b4..1fa8c68d3 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_80154C24.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_80154C24.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
@@ -2137,7 +2137,74 @@ INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006_jr_80154C24", func_8015660
 DEFINE_func_80156648()  /* dedup: shared engine-core @0x80156648 (src/shared) */
 
 
-INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006_jr_80154C24", func_80156670);
+// @class: regalloc-order
+// @stuck: none — MATCH (83 ins)
+//
+// Body notes (what makes the codegen land):
+//   * param_2/param_3 MUST be s32 (not u16) — u16 params flip the $s5/$s6 tie-break and reorder
+//     the prologue; the explicit `& 0xFFFF` at the call site is what yields the two `andi`s.
+//   * D_80126AF0 as `S8[]` (8-byte stride) folds the guard load into
+//     `lui %hi(D_80126AF6)` + `addu $at,$at,$s0` + `lh %lo(D_80126AF6)($at)` (§18 %lo-fold).
+//   * do-while + `D_801151E0[i]` lets gcc derive all three IVs itself
+//     ($s2 = &D_801151E0[i], $s0 = i*8, $s1 = i) and strength-reduce $s2 to an `addiu 4`.
+//
+// TU-plumbing notes (this is the revision that should BANK; the older
+// .run/drafts-sc07006-fresh-* copies match_one-MATCHed but were struct-walled, §23 case 3):
+//   * `S8` / `B8` are ALREADY defined in src/shared/engine_types.h (Phase-22 type-lift), which
+//     engine_core.h includes — a local `typedef ... S8;` is a hard C89 redefinition error in the
+//     TU. They are therefore emitted only under `#ifndef BFM_ENGINE_TYPES_H` so the draft still
+//     compiles standalone under match_one and is a no-op once spliced into the TU.
+//   * The TU's DEFINE_func_801567BC / _80156848 / _80156A88 / _80156B74 expansions already declare
+//     D_801151E0/D_801150E0/D_8011DAD8/D_80128120/D_80128138/D_80126AF0/D_80126730 with exactly
+//     these types, and D_801270A8 as scalar `S8` (not `u8[]`) — declared canonically here so no
+//     "type mismatch with previous external decl" fires. `(s32)&D_801270A8` is byte-identical to
+//     the old `(s32)D_801270A8` on a `u8[]` (both are just the symbol address).
+
+#ifndef BFM_ENGINE_TYPES_H
+
+
+#endif
+
+extern void func_80156B74(s32 param_1, u32 param_2, u8 *param_3);
+extern void func_80157158(s32 a0, u16 a1, u16 a2, s32 a3, s32 a4, s32 a5,
+                          s32 a6, s32 a7, s32 a8, s32 a9, u16 a10, s32 a11, s32 a12);
+extern s32 D_801151E0[];
+extern s32 D_801150E0[];
+extern u8 D_8011DAD8[];
+extern B8 D_80128120[];
+extern B8 D_80128138[];
+extern S8 D_80126AF0[];
+extern u8 D_80126730[];
+extern u8 D_8011520C[];
+extern S8 D_801270A8;
+
+void func_80156670(s32 param_1, s32 param_2, s32 param_3, s32 param_4, s32 param_5, u16 param_6)
+{
+    u32 i;
+
+    func_80156B74(param_1, param_4, (u8 *)(param_1 + 0x1A9));
+    i = 0;
+    do {
+        D_801151E0[i] = 0;
+        if (D_80126AF0[i].d == 0) {
+            D_801151E0[i] = ((s32 (*)(s32, u16, u16, s32, s32, s32, s32, s32, s32,
+                                      s32, u16, s32, s32))func_80157158)(
+                param_1, param_2 & 0xFFFF, param_3 & 0xFFFF,
+                (s32)&D_801270A8,
+                (s32)&D_80126AF0[i],
+                (s32)&D_80128120[i],
+                (s32)&D_80128138[i],
+                (s32)&D_801150E0[i],
+                (s32)&D_8011DAD8[i * 8],
+                param_5,
+                param_6,
+                (s32)&D_8011520C[i],
+                (s32)&D_80126730[i]);
+        }
+        i = i + 1;
+    } while (i < 3);
+}
+
 
 DEFINE_func_801567BC()  /* dedup: shared engine-core @0x801567bc (src/shared) */
 
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_801588CC.c b/src/ov_SC07_006/ov_SC07_006_jr_801588CC.c
index f242d7b29..3bb68759e 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_801588CC.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_801588CC.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_8015C32C.c b/src/ov_SC07_006/ov_SC07_006_jr_8015C32C.c
index fe49bd90b..480e232b3 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_8015C32C.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_8015C32C.c
@@ -137,7 +137,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
@@ -8491,7 +8491,55 @@ s32 func_8017496C(void *a0) {
 DEFINE_func_801749A8()  /* dedup: shared engine-core @0x801749a8 (src/shared) */
 
 
-INCLUDE_ASM("asm/ov_SC07_006/nonmatchings/ov_SC07_006_jr_8015C32C", func_801749C8);
+// @class: regalloc-order
+// @stuck: none — MATCH (pins $s2=D_80126B5E, $s1=D_80126B66 + Buf_c1 struct local; ported from the banked ov_SC01_000 exemplar)
+
+
+
+extern u16 D_80126B5E;
+extern u16 D_80126B62;
+extern u16 D_80126B66;
+
+extern s32  func_80012C6C(s32 a0, s32 a1, s32 a2);
+extern s32 func_80012ABC(s32, s32, s32);
+extern void func_80049CAC(s32 a0, s32 a1);
+extern void func_8012F14C(s32 a0, s32 a1, s32 a2);
+
+
+s32 func_801749C8(s32 param_1)
+{
+    Buf_c1 buf;
+    register s32 iVar5 __asm__("$18");
+    register u32 iVar4 __asm__("$17");
+
+    *(s32 *)(param_1 + 8) = (s16)((s16 (*)(s32, s32, s32))func_80012C6C)((s32)*(s16 *)(param_1 + 8), (s32)*(s16 *)(param_1 + 0xc), 4);
+    *(s32 *)(param_1 + 0x10) = (s16)((s16 (*)(s32, s32, s32))func_80012C6C)((s32)*(s16 *)(param_1 + 0x10), (s32)*(s16 *)(param_1 + 0x14), 4);
+    *(s16 *)(param_1 + 0x18) = ((s16 (*)(s32, s32, s32))func_80012ABC)((s32)*(s16 *)(param_1 + 0x18), (s32)*(s16 *)(param_1 + 0x20), 4);
+    *(s16 *)(param_1 + 0x1a) = ((s16 (*)(s32, s32, s32))func_80012ABC)((s32)*(s16 *)(param_1 + 0x1a), (s32)*(s16 *)(param_1 + 0x22), 4);
+    *(s16 *)(param_1 + 0x1c) = ((s16 (*)(s32, s32, s32))func_80012ABC)((s32)*(s16 *)(param_1 + 0x1c), (s32)*(s16 *)(param_1 + 0x24), 4);
+    *(s16 *)(param_1 + 0x28) = ((s16 (*)(s32, s32, s32))func_80012C6C)((s32)*(s16 *)(param_1 + 0x28), (s32)*(s16 *)(param_1 + 0x2e), 0x10);
+    *(s16 *)(param_1 + 0x2a) = ((s16 (*)(s32, s32, s32))func_80012C6C)((s32)*(s16 *)(param_1 + 0x2a), (s32)*(s16 *)(param_1 + 0x30), 0x10);
+    *(s16 *)(param_1 + 0x2c) = ((s16 (*)(s32, s32, s32))func_80012C6C)((s32)*(s16 *)(param_1 + 0x2c), (s32)*(s16 *)(param_1 + 0x32), 0x10);
+
+    iVar4 = (u16)D_80126B66;
+    *(s32 *)(param_1 + 0x48) = *(s16 *)(param_1 + 0x28) + (iVar5 = (s16)D_80126B5E);
+    *(s32 *)(param_1 + 0x4c) = *(s16 *)(param_1 + 0x2a) + (s16)D_80126B62;
+    *(s32 *)(param_1 + 0x50) = *(s16 *)(param_1 + 0x2c) + (s16)iVar4;
+    func_80049CAC(param_1 + 0x18, (s32)&buf);
+
+    buf.f14 = *(s16 *)(param_1 + 0x28) + iVar5;
+    buf.f18 = *(s16 *)(param_1 + 0x2a) + (s16)D_80126B62;
+    buf.f1c = *(s16 *)(param_1 + 0x2c) + (s16)iVar4;
+    buf.f20 = 0;
+    buf.f22 = 0;
+    buf.f24 = (s16)*(s32 *)(param_1 + 0x10);
+    ((void (*)(s32, s32, s32))func_8012F14C)((s32)&buf, (s32)&buf.f20, (s32)&buf.f28);
+
+    *(s32 *)(param_1 + 0x3c) = buf.f28;
+    *(s32 *)(param_1 + 0x40) = buf.f2a;
+    *(s32 *)(param_1 + 0x44) = buf.f2c;
+}
+
 
 DEFINE_func_80174B6C()  /* dedup: shared engine-core @0x80174b6c (src/shared) */
 
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_8017AE2C.c b/src/ov_SC07_006/ov_SC07_006_jr_8017AE2C.c
index 215d6b2a5..ea7b09a3f 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_8017AE2C.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_8017AE2C.c
@@ -139,7 +139,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/ov_SC07_006/ov_SC07_006_jr_8017BEBC.c b/src/ov_SC07_006/ov_SC07_006_jr_8017BEBC.c
index 77666ee55..aaff64d8e 100644
--- a/src/ov_SC07_006/ov_SC07_006_jr_8017BEBC.c
+++ b/src/ov_SC07_006/ov_SC07_006_jr_8017BEBC.c
@@ -139,7 +139,7 @@ extern s8 D_801152C0;
 extern u8 D_80127504;
 extern void func_800144D4(void);
 extern void func_80129C40(s32 _arg0);
-extern void func_8012A328(void);
+extern void func_8012A328();
 extern void func_80053308(s32);
 extern s32 func_80012F74(s32, s32, s32, s32);  /* canonical s32 (engine_core); (s16)-cast the return for the sll/sra */
 extern void GsSetRefView2L(void *);
diff --git a/src/shared/engine_core.h b/src/shared/engine_core.h
index b7057645d..35c36081c 100644
--- a/src/shared/engine_core.h
+++ b/src/shared/engine_core.h
@@ -2027,7 +2027,7 @@
     }
 
 #define DEFINE_func_801749A8() \
-    extern s32 func_801749C8(s32); \
+    extern s32 func_801749C8(); \
     s32 func_801749A8(s32 a0) { \
         return func_801749C8(a0); \
     }