Drew T
1431b523a4
- recovered the engine actor/player struct cold via PCSX-Redux (R10/R11): base 0x80078E00, ~154 fields over 0x24C bytes, 32 pointer fields; HP/BP/gold/day/hour/position/BP-drain all live-verified and anchored to the Phase-3 player state + the walk diff. -> docs/actor-struct.md - tools/ram_probe.py: PCSX-Redux RAM capture/diff/read/ptr-scan workhorse for field-typing - FINDING (the matching verdict): feeding the struct to m2c as --context is BYTE-NEUTRAL. m2c adopts it (arg0->f_X) but structural-miss sample scores 0 better / 10 same / 2 worse vs m2c-alone (full code+RAM pointer set). The misses are 50-98% mismatched = m2c decompiler limits, not missing types. T4 fn-ptr context unblocks compilation but functions stay structural. - conclusion: rich context (struct/type recovery) is a comprehension / decomp.me win, NOT a byte-match lever -> Phase-16 wall re-confirmed from the type angle. struct-core-pivot.md updated. - emulator field-naming pass deferred (comprehension, off the matching path); NEXT: T6 permuter validation on the 146 genuine near-misses, then T7 go/no-go.
3.1 KiB
Actor / Player struct — recovered layout + the byte-neutral finding (Phase 17 / T4-T5)
Base:
0x80078E00(the player instance). ~154 fields over 0x24C bytes. This is the engine's central entity record —arg0to 550/941 of the struct-heavy residual functions. Recovered by (a) static inference from the union of m2cM2C_FIELD(arg0, T, off)accesses across all drafts, (b) a live PCSX-Redux RAM pass (R10/R11), anchored to the Phase-3 player-state fields.
⚠ Key finding (the matching verdict)
Feeding this struct to m2c as --context is BYTE-NEUTRAL for the structural-miss bucket. Measured: on a
12-function structural-miss sample, m2c-with-struct vs m2c-alone → 0 better / 10 same / 2 worse (even with
the full code-inferred + RAM pointer set, 32 pointers). arg0->f_X and M2C_FIELD(arg0, T*, X) emit identical
bytes; the structural misses (50–98% instruction mismatch) are m2c decompiler limits, not missing types — so
the struct does not crack them, and neither will the permuter at that mismatch magnitude. This confirms the
Phase-17 deep-research caveat: type recovery is comprehension, not a blanket match-% lever. The struct is
kept as a Gen2 quality artifact + decomp.me enabler, not a build input.
Confirmed fields (live-verified, R10)
| offset | type | meaning | evidence |
|---|---|---|---|
| +0x08C | s32 | gold | =126 (Phase-3 anchor) |
| +0x0AC | u8 | day | =2 |
| +0x0B1 | u8 | hour | =22 |
| +0x0B2 | u16 | hp_max | =150 |
| +0x0B4 | u16 | hp_cur | =150 (the plan's "+0xB4"; Phase-3 0x80078EB4) |
| +0x0B6 | u16 | bp_max | =150 |
| +0x0B8 | u16 | bp_cur (stamina) | 74→50 as the player walked (BP-drain) |
| +0x108 | s16 | position X | +6 on a walk north |
| +0x10C | s16 | position Z | +10 on a walk north |
| +0xF0 | s16 | step/heading counter | +1 per walk tick |
Pointer fields (the m2c-can't-infer set; idle RAM scan + code-inferred)
RAM-observed: +0x54 → 0x8017E75C, +0x58 → 0x8017E720, +0x68 → 0x8017E91C (overlay code — likely
behavior/handler function pointers), +0x6C/+0x70 → 0x800D1E20 (resident), +0x5C → 0x8010F5C8. Plus ~26 more
offsets m2c dereferences as pointers in code (e.g. +0x08, +0x4C, +0x50). Full set in .run/actor_fields.json.
Reproduce
- Static layout: aggregate
M2C_FIELD(arg0, T*, off)over.run/wt_drafts/*.c(regen via the harvest). - Live typing:
tools/ram_probe.pyagainst PCSX-Redux (host in.run/ram/HOST.txt); snapshot/diff across controlled actions (walk/jump/swing/charge/damage) to type + name fields (R10 multi-datapoint). - The validation (struct byte-neutral): m2c-with-
.run/actor_ctx2.c→match_onevs m2c-alone.
Deferred (comprehension, not the matching critical path)
Full field-naming via the emulator scenario list (jump=air-state/y-vel; swing=action-state+weapon-id; charge=charge-level; damage=hp/hurt-state; the +0x14/+0x15 dispatch-state per memory-map; the flashing-HUD "interactable nearby" Bincho flag). Resume when comprehension/decomp.me work is prioritized — it does not move the byte-match needle (proven above).