The legacy I2S driver buffers i2s_write() data in a TX DMA ring
(8x64 samples = 64 ms @ 8 kHz) and hands a buffer back to i2s_write()
only after the DMA finished playing it. After the last queued event
the sound task called i2s_zero_dma_buffer(), which memsets the whole
ring instantly — erasing everything not yet played. Any sound <= 64 ms
was written into free ring space without ever blocking, so the zeroing
wiped it before the DMA reached a single sample: completely silent on
hardware. Longer tones force i2s_write() to block on a full ring,
pacing rendering to real-time playback — hence only long beeps were
audible, and the fade-envelope cap never moved the threshold. Every
game beep <= 80 ms (link-active 10, laser 30, sever 40, bleep/ICE 50,
feedback 80) sat in the dead zone. The emulator never showed it: its
i2s_write shim fed SDL directly and had no ring at all.
Fix: replace the instant zeroing with a paced silence purge — once the
queue drains, write one ring + one chunk of zero samples through
i2s_write(), reusing each DMA buffer only after the DMA played it. The
sound plays out fully and the ring then loops silence (still kills the
tail-replay bug the zeroing existed for). Keep the envelope cap at
total/4 so short tones keep a steady-state portion.
Emulator shim gains a TX DMA ring model (write/play cursors + bounded
DAC-emitted log) so the bug class reproduces headlessly: unit/
sound_short_tone_reaches_dac fails on the zero_dma code, passes on the
purge (A/B verified); shim_zero_dma_erases_unplayed pins model fidelity;
sound_long_tone_plays_out_fully guards against truncation. SoundEngine
gains a pumpOne() test seam and begin(spawnTask) for deterministic
driving. Full suite: 26 unit + 7 integration + 4 E2E, 0 failures.
Root cause (on device): sound_engine zero-initialized i2s_pin_config_t,
leaving mck_io_num == 0. i2s_set_pin() then claimed GPIO0 — an
MCLK-capable pin on the S3 and the trackball centre / BOOT line — as
its MCLK output. Verified on device with a diagnostic build: pin0 read
LOW at idle, exactly one latched FALLING edge, clicks never registered.
- sound_engine: set pins.mck_io_num = I2S_PIN_NO_CHANGE explicitly.
- NetDeck: register the trackball click on release with contact-bounce
collapse (the proven CrtTerminal pattern); pollTouch re-arms when
isPressed() stays true but getPoint() reports no points (stuck GT911
IRQ on some units).
- Emulator parity (so this class of bug dies before flashing):
GPIO-matrix pin-ownership model in the shim — i2s_set_pin() claims
its pins, claimed pins are dead inputs (digitalRead LOW, no edges,
loud WARNING on claim/ISR collisions); host pin writes go through
emu::setPin/togglePin with an edge-accurate ISR latch (a full
press+release pulse between dispatches no longer gets swallowed);
E2E inits SDL_INIT_AUDIO so the I2S path actually runs there.
Regression coverage: e2e/trackball_click (fails on the old engine),
unit/shim_claimed_pin_is_dead_input,
unit/shim_edge_survives_same_dispatch_pulse.
- Makefile: firmware step follows the sketch move (root platformio.ini,
src_dir = src/netdeck).
- HW.md: new notes file for emulator-vs-hardware quirks and the
on-device debugging recipe.
On-device verified: clicks register in both views (beacon drops, node
routing). Fast layer 23 unit + 7 integration green. Note: e2e
boot_and_controls + network_journey failures are pre-existing at HEAD
(reproduced on a clean stash before this change).
- Replace hardcoded with in NetDeck.ino (8 occurrences)
- Add to nd_layout.h
- Update test expectations in unit_tests.h and integration_tests.h from units[6] to units[7] to match the new reinforcement slot indexing
This enables supporting more than 7 units in the future.
Replace the fading visibility[] stub with two bitmaps over the tile grid:
fogNow (visible this tick) and fogSeen (persistent historical memory),
re-derived each sim tick via revealLos() Bresenham raycasts from:
- each living marine's 8-tile sight window (= marine fire range, so the
squad never shoots into darkness), and
- each online network node uncovering its connected tiles (3-tile window).
Rendering (renderMap): never-seen -> flat dark gray, historical ->
luma grayscale via gray565(), current -> full palette. Hostiles on
non-visible tiles are hidden in renderUnits. Ad-hoc revealTile() stamps
(marine step, door/HIVE links, hand-placed airlock tiles) removed; fog
refreshes on door toggle and scenario reset.
Tests: 5 new unit tests (initial reveal, wall blocking with face render,
historical persistence, node uncover/sever, gray565 channels) plus a
regression pinning fog sight >= marine fire range.