feat(fds,input): FDS medium model + Famicom mic + Zapper aperture (v2.2.0 Capstone)#291
Merged
Conversation
Contributor
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
There was a problem hiding this comment.
Pull request overview
This PR implements the v2.2.0 “Capstone” FDS medium-model capstone (wire-image synthesis with per-block CRC re-emission on write, plus an opt-in analog head-seek model) and completes the remaining input peripherals (Famicom microphone + Zapper aperture-based light detection), with corresponding docs/changelog updates. The changes are intended to remain additive/default-off so existing default execution stays byte-identical.
Changes:
- FDS medium-model completion: synthesized wire image with per-block CRC-16/KERMIT, CRC re-emitted on write, synthetic write-verify oracle, and additive v4 save-state tail for analog head-seek state.
- Input additions/hardening: Famicom microphone surfaced on
$4016.D2end-to-end (frontend → core), and Zapper light detection upgraded to a 3×3 aperture with a minimum-bright-pixel threshold. - Documentation updates across mapper docs, frontend docs, the accuracy ledger, and the changelog.
Reviewed changes
Copilot reviewed 12 out of 12 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| docs/mappers.md | Adds documentation for the new FDS medium model behavior and verification approach. |
| docs/frontend.md | Documents the new “Capstone” peripherals (mic, keyboard note, Zapper aperture sampling). |
| docs/accuracy-ledger.md | Updates the ledger entries to reflect the shipped FDS medium model + mic + Zapper changes and their verification split. |
| crates/rustynes-mappers/src/fds.rs | Implements CRC re-synthesis on writes, opt-in analog head-seek timing, synthetic medium verification, and bumps FDS save-state tail to v4. |
| crates/rustynes-frontend/src/input.rs | Tracks a hold-to-talk mic key and exposes it via InputState. |
| crates/rustynes-frontend/src/emu.rs | Adds FrameInputs.microphone and forwards it to Nes::set_microphone. |
| crates/rustynes-frontend/src/emu_thread.rs | Extends SharedInput to carry the mic state across threads. |
| crates/rustynes-frontend/src/app.rs | Wires mic state into FrameInputs on native builds. |
| crates/rustynes-core/src/nes.rs | Adds Nes::set_microphone API. |
| crates/rustynes-core/src/input_device.rs | Implements Zapper 3×3 aperture light sampling + tests. |
| crates/rustynes-core/src/bus.rs | Adds mic state to bus and ORs it into $4016 reads + tests. |
| CHANGELOG.md | Documents the user-visible v2.2.0 Capstone features. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
doublegate
added a commit
that referenced
this pull request
Jul 12, 2026
…+ FDS tail offset invariant Two Copilot review threads on the v2.2.0 "Capstone" FDS + peripherals PR, both valid, both fixed with default-off byte-identity preserved. 1. bus.rs — DMC-DMA register-conflict read of $4016 dropped the microphone bit. The normal $4016 read path composes D2 from `famicom_mic` (`<< 2`), but `dmc_dma_read`'s `conflict_addr == 0x4016` branch rebuilt the byte from `controllers[0].read()` alone, so a DMC-DMA that collides with a $4016 read while the mic is asserted would have incorrectly cleared bit 2. Now OR in the same `mic = u8::from(self.famicom_mic) << 2` term, keeping the conflict path consistent with the normal controller-read semantics. Mic is default -off (released ⇒ mic == 0), so the returned byte is byte-identical to prior releases on the default; only an actively-held Famicom mic during a DMC/OAM DMA overlap is affected — the accurate behavior. 2. fds.rs — `load_disk_tail` read the v4 `pre_rewind_head` u32 out of `data[off..off+4]` without advancing `off`, breaking the "offset always reflects consumed bytes" invariant (functionally inert today since `off` was dead afterward, but a hazard for any future tail extension). Advance `off += 4` after the read, and add a `debug_assert_eq!(off, data.len())` that both documents/enforces the invariant (the top-of-fn length check already proved `expected == data.len()`, and both the v4 and v3 paths consume exactly to the blob end) and reads `off` on every path, so the new final assignment does not trip `unused_assignments` under `-D warnings`. Gate: fmt/clippy (workspace + frontend scripting/hd-pack/retroachievements)/ no_std thumbv7em/rustdoc/markdownlint all green; AccuracyCoin 141/141 (100%), visual_regression + nestest + pal_apu_tests byte-identical, FDS synthetic write-verify + NES-level save-state round-trip pass. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
doublegate
force-pushed
the
feat/v2.2.0-fds-peripherals
branch
from
1b5959c to
9dcdb3d
Compare
…perture Emulation capstone (v2.2.0). FDS medium model: continuous belt-velocity head-seek model (opt-in, default-off, byte-identical fixed-window default), per-block CRC-16 re-emission on write, a BIOS-free synthetic write-verify oracle (medium_write_verify), and an additive v4 save-state tail. Peripherals: Famicom controller-2 microphone modeled on $4016 bit 2 (default-off, byte-identical), and a 3x3 photodiode-aperture Zapper light-sample replacing the single-pixel model (deterministic, no save-state change). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…apstone) Thread the Famicom mic (M key) through InputState -> FrameInputs -> SharedInput -> EmuCore latch (Nes::set_microphone). Docs: mappers.md FDS medium model, accuracy-ledger.md CI-verifiable-vs-local-only split, frontend.md peripherals, CHANGELOG [Unreleased]. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Convert doc links to pub(crate)/private items (FdsMediumError, reseek_window_cycles, ZAPPER_APERTURE_*) to plain code spans so the default cargo doc build passes -D warnings (project convention for feature-/visibility-limited item names).
…+ FDS tail offset invariant Two Copilot review threads on the v2.2.0 "Capstone" FDS + peripherals PR, both valid, both fixed with default-off byte-identity preserved. 1. bus.rs — DMC-DMA register-conflict read of $4016 dropped the microphone bit. The normal $4016 read path composes D2 from `famicom_mic` (`<< 2`), but `dmc_dma_read`'s `conflict_addr == 0x4016` branch rebuilt the byte from `controllers[0].read()` alone, so a DMC-DMA that collides with a $4016 read while the mic is asserted would have incorrectly cleared bit 2. Now OR in the same `mic = u8::from(self.famicom_mic) << 2` term, keeping the conflict path consistent with the normal controller-read semantics. Mic is default -off (released ⇒ mic == 0), so the returned byte is byte-identical to prior releases on the default; only an actively-held Famicom mic during a DMC/OAM DMA overlap is affected — the accurate behavior. 2. fds.rs — `load_disk_tail` read the v4 `pre_rewind_head` u32 out of `data[off..off+4]` without advancing `off`, breaking the "offset always reflects consumed bytes" invariant (functionally inert today since `off` was dead afterward, but a hazard for any future tail extension). Advance `off += 4` after the read, and add a `debug_assert_eq!(off, data.len())` that both documents/enforces the invariant (the top-of-fn length check already proved `expected == data.len()`, and both the v4 and v3 paths consume exactly to the blob end) and reads `off` on every path, so the new final assignment does not trip `unused_assignments` under `-D warnings`. Gate: fmt/clippy (workspace + frontend scripting/hd-pack/retroachievements)/ no_std thumbv7em/rustdoc/markdownlint all green; AccuracyCoin 141/141 (100%), visual_regression + nestest + pal_apu_tests byte-identical, FDS synthetic write-verify + NES-level save-state round-trip pass. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
doublegate
force-pushed
the
feat/v2.2.0-fds-peripherals
branch
from
9dcdb3d to
51f2412
Compare
doublegate
added a commit
that referenced
this pull request
Jul 12, 2026
…ripherals + quality/security) (#297) The milestone cut that closes the v2.1.5 -> v2.2.0 "deepen the existing project" run, landing its two remaining marquees plus a peripherals and quality/security pass. Every change already merged (PRs #290, #291) is additive or default-off; the deterministic core is untouched, so this cut itself moves only version strings + release docs and no framebuffer/audio byte. Marquees (already on main via #290/#291): - Netplay matchmaking + lobby (B5) — a browse-and-join room directory (`ListRooms`/`RoomList`) and server-side quick-play (`QuickMatch`/`Matched`) over the existing room-code / TURN transport, delayed-stream spectators (`SpectatorConfig.delay_frames`, reveal-horizon gated), a graded hysteresis desync verdict (`DesyncStatus`, threshold 3 ~= 1.5 s), and graded multi-second peer-liveness RTT timeouts (`PeerLink` + `DisconnectReason::PeerTimeout`). Signaling and telemetry only — the rollback/determinism contract is untouched. - FDS medium model completion (F4.3) — per-block CRC-16/KERMIT re-emitted on every BIOS write (`resynth_block_crc`) over a synthesized gap/mark wire image, an opt-in default-OFF continuous analog head-seek / velocity model (`Fds::set_analog_head_seek`) replacing the flat `HEAD_RESEEK_CYCLES` window, and a BIOS-free synthetic write-verify oracle (`Fds::medium_write_verify`) as the CI-verifiable half (the real-BIOS write-CRC path stays gitignored/local — copyright disksys.rom). Additive v4 FDS save-state tail; with the model off a non-writing `.fds` run is byte-identical to prior releases. Peripherals (#291): the Famicom `$4016`-bit-2 microphone (`Nes::set_microphone`, `$4016`-only, never `$4017`) and a 3x3-aperture Zapper light-timing model (`ZAPPER_APERTURE_*`), both additive/default-off. Quality/security (#290): cargo-fuzz targets grown 3 -> 8 (`ppu_reg_io`, `apu_reg_io`, the high-value `netplay_message`, `save_state`, `movie`), which surfaced and fixed two real OOM-DoS paths in `Movie::deserialize` (untrusted `frame_count` -> unbounded `Vec::with_capacity`, and a zero-`bytes_per_frame` infinite push loop) byte-identically for valid input; a read-only Tools -> ROM Info browser (dump-identity CRC32 keys + SHA-256 + per-game DB entry + decoded header, `&Nes`, never mutates); and four new MkDocs handbook pages. Cut mechanics: `[workspace.package] version` 2.1.10 -> 2.2.0 (Cargo.lock refreshed across all 18 workspace crates), CHANGELOG `[Unreleased]` renamed to `[2.2.0]` with a fresh empty `[Unreleased]`, `.github/release-notes/ v2.2.0.md` authored, and README badge / BibTeX / current-release paragraph updated. `release-auto.yml` tags + publishes on main going green. Sacred gate (verified on the merged #290+#291 tree): AccuracyCoin 141/141 (100.00%), nestest 0-diff, `pal_apu_tests` 10/10, `visual_regression` byte-identical, netplay rollback-determinism + snapshot-restore-replay + FDS synthetic write-verify green, `no_std` thumbv7em cross-compile, full host fmt/clippy/rustdoc/markdownlint, version consistency. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The emulation capstone for RustyNES v2.2.0 "Capstone": the FDS medium
model (F4.3, the marquee) plus the remaining input peripherals. All changes are
additive and default-off, so the shipped default (non-writing
.fdsruns +the standard controller path) stays byte-identical and AccuracyCoin holds
141/141 (100%).
FDS medium model (F4.3, marquee) —
crates/rustynes-mappers/src/fds.rsmedium (lead-in / inter-block gaps,
$80start mark, block bytes, aCRC-16/KERMIT per block). Each BIOS-written block now re-emits a fresh
per-block CRC-16 over its updated payload (
resynth_block_crc), modellingthe RP2C33 controller's continuous CRC generator so the medium stays
self-consistent after a write.
Fds::set_analog_head_seek). Replaces the flat fixedHEAD_RESEEK_CYCLESmotor-restart not-ready window with a belt-driven, distance-proportional
seek time (
HEAD_SEEK_BYTES_PER_CYCLEvelocity +HEAD_SEEK_SETTLE_CYCLESsettle, clamped to a cold spin-up), sized from the head-travel distance
captured at motor-off. With the model off (the default), a non-writing
.fdsrun is byte-identical.
medium_write_verify). A BIOS-free walkof the wire image asserting every block's CRC-16 and gap/mark framing
round-trips — the CI-verifiable half of the medium model. The real-BIOS
write-CRC path needs a copyright
disksys.romand is exercised only from agitignored
tests/roms/external/dump (documented indocs/accuracy-ledger.md).load with the model disabled.
Remaining peripherals
$4016bit 2 (Nes::set_microphone/
Bus::set_microphone), wired through the frontend input path (hold-to-talkMkey ->FrameInputs.microphone->SharedInput-> latch). A$4016-onlysignal (never
$4017); default-off leaves$4016byte-identical. ForZelda (Pols Voice), Kid Icarus, etc.
9x8matrix(
FamilyKeyboardState+ Subor clone) and its frontend key mapping werealready modelled; verified complete, unchanged here.
(field-of-view) around the aim point, asserting light only when >=2 pixels
cross the luma threshold (
ZAPPER_APERTURE_*), hardening detection vs theprior single-pixel sample. Deterministic (pure function of the presented
framebuffer), no save-state change. The finer ~19-26-scanline temporal hold is
below per-frame sample resolution (supported titles re-poll every frame); full
per-dot temporal integration is a documented future refinement.
Verification
visual_regression/nestest/pal_apu_tests(20 total);save_state+battery_save(12); FDS mapperunit tests (69, incl. the new synthetic write-verify + head-seek + v4
round-trip);
input_devicesharness (8); core+mappers (855); frontend lib(450, incl. the mic keybind test).
cargo fmt --check;clippy --workspace --all-targets -D warnings+ frontendscripting/scripting,hd-pack/retroachievementscombos;
RUSTDOCFLAGS=-D warnings cargo doc; markdownlint (pinned v0.39.0);no_stdthumbv7em-none-eabihfcross-compile.Docs
docs/mappers.md(FDS medium model),docs/accuracy-ledger.md(CI-verifiablesynthetic vs local-only real-BIOS),
docs/frontend.md(Capstone peripherals),CHANGELOG.md[Unreleased].Do NOT merge — the maintainer runs the release ceremony.
🤖 Generated with Claude Code