[DrLegitamate]

Follow-up to #394: two Oozla (level 01) functions matched from their retained near misses, and an improved near miss for Joba's largest remaining function.

Matched

Function	Bytes	What closed it
func_L01_003D1A48	1,268	The RGB mask is a block-local int rgb = 0xFFFFFF; at the top of the inner loop body. As an inline constant its lifetime was 1, so loop.c's first pass (threshold already lowered by two float hoists) left it for the rerun, after five pointer preparations; as a local its set is the first movable insn and is hoisted first, giving retail's preheader order. -O2 -G8 -g in its own unit.
func_L01_003D8358	1,108	Two source corrections. step = speed*step swapped the mul.s operands (expand_binop swaps them when the target is an operand); the height now adds speed*step directly. Reusing scale for both the constant and the product tied the scale multiply with an argument move in sched2 (equal priority, decided by dependent count); holding the constant in its own k gives retail's order. Default -G8 profile; its port-only C row and file are removed (as in 588f8f0).

Notes only

• func_L11_00420C88 (15,832 bytes): the retained body is now 15,832/15,832 bytes with 18 register-only rows left. Earlier drafts' counts hid 47 alignment nops from a do/while(0) macro. Recovered: per-branch call tails, retail branch order, nested call arguments, strict float compares, case-local variables, the case 5 partner loop in the shape of matched sibling func_L11_00424A60 (its D_L11_001B359C load now lands in the branch slot as retail's one-word $gp form), and a real index error in case 19 (0x54c where retail uses 0x56c).

Question for maintainers: func_002B60B0's declaration in Joba callers

Two independent Joba functions match completely only if this caller declares func_002B60B0 as returning void:

• func_L11_00420C88: with the retained body, a void declaration closes all 18 remaining rows (verified: 1 of 1 places match). With the true void * return, sched1 ranks the post-call index load above the lui and local-alloc sees $v0 as still in use, so index and base swap $v0/$v1 at three call sites.
• func_L11_00452450: a void declaration closes its seven rows (see its note).

src/game/002B60B0.c returns its pointer, and COMPILER-NOTES forbids an invented void return, so neither is used here. Most existing callers in src/ already declare it void. If maintainers accept that these callers' original declaration was void (retail leaving $v0 free after the call at every site is consistent with that), both functions can land as matches; otherwise the notes stand.

Validation

• tools/check.py: each matched function 1 of 1 USA v1.01 place. func_L01_003D8358 also has a USA v2.00 place, not checked here (no US2 dump on this machine); PAL has neither.
• tools/build.py all --strict: 28 programs OK, zero C fallbacks (01_oozla: OK: 2062852 loaded bytes equal).
• python3 -m unittest discover -s tests: 339 tests OK.
• tests/check_repository.py: repository checks passed.
• Progress report not regenerated.

🤖 Generated with Claude Code

https://claude.ai/code/session_017f7s8MnhdrdEynhYhxkLs6

[bytematchdev]

ByteMatch progress report

SCES_516.07

Code matched	54.15% · 3.09 MB of 5.70 MB
Fuzzy match	54.15%
Functions	6,996 / 8,672
Data matched	0.00%
Units complete	958 / 1,196

No function changes.

SCUS_972.68_v2

Code matched	81.01% · 4.52 MB of 5.58 MB
Fuzzy match	81.01%
Functions	7,595 / 8,634
Data matched	0.00%
Units complete	973 / 1,156

No function changes.

SCUS_972.68

Code matched	84.74% · 4.68 MB of 5.53 MB
Fuzzy match	84.74%
Functions	7,760 / 8,579
Data matched	0.00%
Units complete	924 / 1,018

No function changes.

_{Commit 50a05a1 · Ratchet & Clank: Going Commando on ByteMatch}