Gatchan/FF6Portage

Watch

Fork 0

RAM-execution trace and inventory; route narshe_to_world_map (refs #8) #23

Merged

Gatchan merged 6 commits from ram-exec into main

2026-10-29 16:47:09 +00:00

Gatchan commented

2026-10-29 16:43:59 +00:00

Owner

Inventory of the code FF6 executes from RAM, measured with a trace built into the core, on a route library that now reaches the first world map. US v1.0 and JP v1.0.

Changes

RAM-execution trace in the debug API (ff6_debug_ram_trace_*, core/port/ram_trace.c), fed by the single instruction-hook dispatcher of core/port/port.c (ports first, then the trace). Exposed as ff6cli --trace-ram <tsv> and the drive-server trace start|stop|write. Usage and format: docs/ram_trace.md.
tools/ram_trace_report.py: merges addresses into ranges, resolves them with the .dbg of the version, exits 3 when code ran outside the RAM map [[code]] images.
RAM map: route variables for the Narshe escape (targetable monsters, max HP, Magitek cursor, world brightness, event flags, world_party_pixel). ram_map_check 0 errors.
New route narshe_to_world_map (mines, Whelk, Arvis, Terra's flight, moogle defence, first world map with the player in control), with FieldGrid navigation from tile properties. Movies US/JP and ctests route/movie/route_ports_on.

Measured (committed movies, from power-on)

Route	Frames US / JP	NMIs via `$1500` US / JP	IRQs via `$1504` US / JP
title_to_first_battle_win	10 253 / -	9 458 / -	7 670 / -
field_menu_tour	10 897 / 10 202	10 008 / 9 315	7 702 / 7 326
narshe_to_world_map	79 761 / 80 670	75 687 / 76 477	37 224 / 35 474

Mode	NMI US / JP	IRQ US / JP
Field	FieldNMI `$C0:0182` / same	FieldIRQ `$C0:0262` / same
Title, opening	CutsceneNMI `$7E:7233` / `$7E:71BF` (RAM)	not entered
Battle	BattleNMI `$C1:0BA7` / `$C1:0B8A`	not entered
Menu (field menu, name entry)	MenuNMI `$C3:1387` / `$C3:13F2`	not entered
World map, walking	WorldNMI `$EE:A728` / `$EE:A71F`	WorldIRQ `$EE:A93A` / `$EE:A931`

RAM code executed: only the trampolines $00:1500/$1504 and the cutscene program ($7E:5000-7B42 US, $7E:5000-7AE4 JP). 0 unexpected addresses in every mode reached, so no finding.

Out of reach (The End, ending, World of Ruin cutscene, vehicles and their raster IRQs, the battle-exit rtl): #22, which proposes late-game start points (route-produced SRAM or an SRAM synthesiser).

Validation

Builds without warnings: debug, clang, mingw (--no-test); release --fast: 13/13.
ctest -L slow (release): 21/21 passed, including route/movie/route_ports_on of narshe_to_world_map on US and JP (about 15 min each).

Refs #8 (the remaining part is in #22)

🤖 Generated with Claude Code

Inventory of the code FF6 executes from RAM, measured with a trace built into the core, on a route library that now reaches the first world map. US v1.0 and JP v1.0. ## Changes - **RAM-execution trace** in the debug API (`ff6_debug_ram_trace_*`, `core/port/ram_trace.c`), fed by the single instruction-hook dispatcher of `core/port/port.c` (ports first, then the trace). Exposed as `ff6cli --trace-ram <tsv>` and the drive-server `trace start|stop|write`. Usage and format: `docs/ram_trace.md`. - `tools/ram_trace_report.py`: merges addresses into ranges, resolves them with the `.dbg` of the version, exits 3 when code ran outside the RAM map `[[code]]` images. - RAM map: route variables for the Narshe escape (targetable monsters, max HP, Magitek cursor, world brightness, event flags, `world_party_pixel`). `ram_map_check` 0 errors. - New route **`narshe_to_world_map`** (mines, Whelk, Arvis, Terra's flight, moogle defence, first world map with the player in control), with `FieldGrid` navigation from tile properties. Movies US/JP and ctests `route`/`movie`/`route_ports_on`. ## Measured (committed movies, from power-on) | Route | Frames US / JP | NMIs via `$1500` US / JP | IRQs via `$1504` US / JP | |---|---|---|---| | title_to_first_battle_win | 10 253 / - | 9 458 / - | 7 670 / - | | field_menu_tour | 10 897 / 10 202 | 10 008 / 9 315 | 7 702 / 7 326 | | narshe_to_world_map | 79 761 / 80 670 | 75 687 / 76 477 | 37 224 / 35 474 | | Mode | NMI US / JP | IRQ US / JP | |---|---|---| | Field | FieldNMI `$C0:0182` / same | FieldIRQ `$C0:0262` / same | | Title, opening | CutsceneNMI `$7E:7233` / `$7E:71BF` (RAM) | not entered | | Battle | BattleNMI `$C1:0BA7` / `$C1:0B8A` | not entered | | Menu (field menu, name entry) | MenuNMI `$C3:1387` / `$C3:13F2` | not entered | | World map, walking | WorldNMI `$EE:A728` / `$EE:A71F` | WorldIRQ `$EE:A93A` / `$EE:A931` | RAM code executed: only the trampolines `$00:1500/$1504` and the cutscene program (`$7E:5000-7B42` US, `$7E:5000-7AE4` JP). **0 unexpected addresses** in every mode reached, so no finding. Out of reach (The End, ending, World of Ruin cutscene, vehicles and their raster IRQs, the battle-exit `rtl`): #22, which proposes late-game start points (route-produced SRAM or an SRAM synthesiser). ## Validation - Builds without warnings: debug, clang, mingw (`--no-test`); `release --fast`: 13/13. - `ctest -L slow` (release): **21/21 passed**, including route/movie/route_ports_on of `narshe_to_world_map` on US and JP (about 15 min each). Refs #8 (the remaining part is in #22) 🤖 Generated with [Claude Code](https://claude.com/claude-code)

Gatchan added 6 commits

2026-10-29 16:44:00 +00:00

tools: RAM-execution trace in the debug API, ff6cli --trace-ram and drive-server trace command (refs #8 ) 7467c09303

One instruction-hook dispatcher in port.c: the port registry runs first, then the
trace records the instruction that really executes (one flag test when off).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

state: RAM map route vars for the Narshe escape: targetable monsters, max HP, Magitek cursor, world brightness and event flags (ram_map_check 0 errors, 161 JP checks) fa4f4ee731

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

state: world_party_pixel, the world map position in pixels (ram_map_check 0 errors, 162 JP checks) 08bc7346ff

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

verify: route narshe_to_world_map to the first world map, field navigation from tile properties, movies US/JP, ctest route/movie/route_ports_on (refs #8 ) d318f968eb

US 79761 frames, JP 80670 frames from power-on (start: end of title_to_first_battle_win).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tools: ram_trace_report.py (ranges and transitions with labels, code outside the RAM map images flagged); serve test covers the RAM trace (refs #8 ) 1c4e0be1e0

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

docs: ram_trace.md, RAM-execution trace usage, TSV format and measured handlers per mode US/JP (refs #8 ) 468362b7b4

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Gatchan merged commit 9ab2982b7b into main

2026-10-29 16:47:09 +00:00

Gatchan referenced this pull request

2026-10-29 16:47:11 +00:00

Complete the RAM-executed code inventory: menu, world map, vehicles, The End #8

Gatchan referenced this pull request from a commit

2026-10-29 16:47:13 +00:00

Merge pull request 'RAM-execution trace and inventory; route narshe_to_world_map (refs #8)' (#23) from ram-exec into main

No reviewers

No labels

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

Gatchan/FF6Portage!23

No description provided.

Rows
Columns