[Clank700]

What

99 NTSC-U (SCUS_974.65) functions move from unmatched to matched, one C file each; 12 of them replace their near miss in nonmatching/ (file and README row removed). US matched functions 5,435 → 5,534 (414,340 → 421,364 bytes), PAL 6,032 → 6,084 (497,616 → 501,228 bytes). config/us/map.tsv, config/pal/map.tsv, both progress reports and the README table are regenerated with the repository tools; only the rows of the new sources changed (our local libm rows, which differ because build/libm was not rebuilt, were restored to main's values).

Every function here is new on both discs: none has a src/pal/ source, so no existing source or its credit moves (no function name appears in two units with different percentages in config/us/map.tsv or config/pal/map.tsv, on main or here). Functions already matched on main, in open PRs (#71, #92, #93 to #97), PAL twins of existing src/pal/ sources and anything that would add an audit_matches.py or audit_relocs.py entry were left out on purpose.

Compiler: 19 of the 99 files carry their own flags

80 files build with the default route (SN GCC 2.95.3, -O2 -G8 -fopt-stack). The other 19 need a per-file header, as the repository already does elsewhere (tools/build.sh reads /* cflags: ... */ from the first lines; nothing in the build scripts changes):

• 5 files: @ee29, the Sony EE GCC 2.9 route (4 × @ee29, 1 × @ee29 -O2 -G0). Part of the library-like and glue code was built with Sony's ee-gcc 2.9 rather than the SN compiler. The repository already supports this whole-file route (commit ddfe94d, tools/cc.sh, docs/TOOLCHAIN.md "Sony EE GCC 2.9 route"). These functions start with sd saves of callee-saved registers at ascending 16-byte spacing, which is the Sony 2.9 prologue style (SN 2.95.3 with -fopt-stack saves with 8-byte slots).
• 5 files: -fno-schedule-insns variants (3 × -fno-schedule-insns -fno-schedule-insns2, 2 × -fno-schedule-insns), which the repository already uses in other sources.
• 4 files -mno-split-addresses (+1 combined with -G0), 1 -G0, 1 -fno-opt-stack, 1 -ffast-math, 1 redundant -O2 -G8 (the default; harmless).

Matching under a per-file profile shows the code is compatible with that profile; it does not prove the historical flag set. The header states only what reproduces the bytes.

How the C was found

Reconstructed from the retail code of this repository (our assembly, m2c) and made to match by compiling with the file's own flags, then checked function by function. Plain C only: no inline assembly, register declarations, M2C_ macros or do { } while (0) shaping, and no source adapted from another project. Declarations of callees were aligned to the existing definitions where that leaves the compiled object byte-identical; three of the new definitions differ in return type from declarations in older files that were not touched (that kind of mismatch already exists on main for about 5% of the definitions).

Checks (clean build of every source of this exact head, on top of main c78d917)

• Build of all 6,340 sources (outside src/libgcc and src/libm) with each file's own flags; build/libgcc and build/libm come from an earlier full build (src/ changes do not affect them).
• tools/audit_matches.py: exit 0, 4,021 exact, 3 not exact (the same 3 as on main: func_00161890, func_00547A98, func_L01_004F3100).
• tools/audit_relocs.py (with the overlay ELFs): exit 1, 80 mismatches in 43 functions, 3 unresolved, identical to main (line by line against a build of 60818dd; only the count of checked functions differs). CONTRIBUTING asks for 0, so we do not call this gate clean; those entries are in sources this PR does not touch.
• tools/match_map.py --check and tools/progress.py --check, for US and --version pal: all exit 0.
• Beyond the repository tools, a per-placement check of the new functions (every catalogued level placement: code bytes with relocation fields masked, and every call target resolved at that placement's own level): no byte difference and no call-target mismatch at any placement that could be checked; negative controls (bytes of another function, call targets shifted by +4) are all rejected.
• Overlap with other open PRs: Match 57 NTSC-U functions #92 (57 NTSC-U functions) also edits the maps, reports and README, so whichever of the two lands second needs tools/match_map.py --write / tools/progress.py run again.

🤖 Generated with Claude Code

[bytematchdev]

ByteMatch progress report

SCES_532.85

Code matched	8.62% (+0.06%) · 501.23 KB of 5.81 MB
Fuzzy match	9.10% (+0.05%)
Functions	6,084 (+52) / 18,113
Units complete	3,282 (+12) / 3,341

✅ Newly matched (72)

• func_L00_0051B458 (264 bytes)
• func_004E8BD8 (224 bytes)
• func_L00_00673950 (192 bytes)
• func_00148800 (180 bytes)
• func_005014E0 (180 bytes)
• func_00502150 (180 bytes)
• func_L01_005A9BF0 (176 bytes)
• func_00425150 (164 bytes)
• func_004E87F8 (164 bytes)
• func_004D3708 (156 bytes)
• func_004A0818 (152 bytes)
• func_004B8F28 (152 bytes)
• func_004BE4F0 (144 bytes)
• func_004B1800 (140 bytes)
• func_004A8A28 (136 bytes)
• func_004B4CE8 (136 bytes)
• func_L00_00519550 (136 bytes)
• func_00473078 (128 bytes)
• func_004E9020 (112 bytes)
• func_L00_00538758 (112 bytes)
• func_L01_00435C00 (104 bytes)
• func_0050F2A0 (96 bytes)
• func_L00_0077A8A8 (96 bytes)
• func_0045EA10 (88 bytes)
• func_00502BB8 (88 bytes)
• func_L01_0041CB10 (84 bytes)
• func_L01_0046A340 (84 bytes)
• func_L41_003E3428 (84 bytes)
• func_0045FC28 (80 bytes)
• func_00464878 (80 bytes)
• …and 42 more

⚠️ No longer matching (20)

• func_004E8BD8 100.00% → 0.00%
• func_005014E0 100.00% → 0.00%
• func_00502150 100.00% → 0.00%
• func_004E87F8 100.00% → 0.00%
• func_004D3708 100.00% → 0.00%
• func_004A0818 100.00% → 0.00%
• func_004B8F28 100.00% → 0.00%
• func_004BE4F0 100.00% → 0.00%
• func_004B1800 100.00% → 0.00%
• func_004A8A28 100.00% → 0.00%
• func_004B4CE8 100.00% → 0.00%
• func_0045EA10 100.00% → 0.00%
• func_00502BB8 100.00% → 0.00%
• func_0045FC28 100.00% → 0.00%
• func_00464878 100.00% → 0.00%
• func_0053DDA0 100.00% → 0.00%
• func_0045CF60 100.00% → 0.00%
• func_0049CF38 100.00% → 0.00%
• func_0049CF80 100.00% → 0.00%
• func_00498590 100.00% → 0.00%

📉 Worse (9)

• func_00148800 88.89% → 0.00%
• func_00473078 89.06% → 0.00%
• func_01EEFE4C 91.67% → 0.00%
• func_01EACB58 86.67% → 0.00%
• func_01ED2DC0 88.46% → 0.00%
• func_01F07D98 87.50% → 0.00%
• func_01F07E5C 87.50% → 0.00%
• func_01F07F20 87.50% → 0.00%
• func_01E9B240 90.00% → 0.00%

SCUS_974.65

Code matched	7.29% (+0.12%) · 421.36 KB of 5.78 MB
Fuzzy match	7.83% (+0.11%)
Functions	5,534 (+99) / 18,085
Units complete	2,872 (+33) / 2,932

✅ Newly matched (99)

• func_L00_0051AF58 (264 bytes)
• func_004E7CB0 (224 bytes)
• func_L00_006702E0 (192 bytes)
• func_00148800 (180 bytes)
• func_00500470 (180 bytes)
• func_005010E0 (180 bytes)
• func_L01_005A5878 (176 bytes)
• func_00424F40 (164 bytes)
• func_004E78D0 (164 bytes)
• func_004D27F8 (156 bytes)
• func_0049FAE0 (152 bytes)
• func_004B8038 (152 bytes)
• func_004BD6F0 (144 bytes)
• func_004B0970 (140 bytes)
• func_004A7CB8 (136 bytes)
• func_004B3E20 (136 bytes)
• func_L00_00519050 (136 bytes)
• func_00472798 (128 bytes)
• func_004E80F8 (112 bytes)
• func_L00_00538258 (112 bytes)
• func_L01_00434DB0 (104 bytes)
• func_L41_00625FE0 (104 bytes)
• func_0050E208 (96 bytes)
• func_L00_007754D0 (96 bytes)
• func_0045E1F8 (88 bytes)
• func_00501B48 (88 bytes)
• func_L01_0041BF10 (84 bytes)
• func_L01_00468F58 (84 bytes)
• func_L01_0064E8F0 (84 bytes)
• func_L41_003E2BF0 (84 bytes)
• …and 69 more

📉 Worse (10)

• func_00148800 88.89% → 0.00%
• func_00472798 89.06% → 0.00%
• func_01EEFE4C 91.67% → 0.00%
• func_01EACB58 86.67% → 0.00%
• func_01ED2DC0 88.46% → 0.00%
• func_01F07D98 87.50% → 0.00%
• func_01F07E5C 87.50% → 0.00%
• func_01F07F20 87.50% → 0.00%
• func_01E9B240 90.00% → 0.00%
• func_00497938 89.29% → 0.00%

_{Commit 1843712 · Ratchet: Deadlocked on ByteMatch}