fix(autoresearch/lpc): move pinToggleRx + ws2812SctTest edge buffers off the stack (closes #3125)

[zackees]

Summary

Closes #3125. Moves the multi-KB edge buffers from stack to right-sized static storage so the LPC845-BRK LowMemory build doesn't stack-overflow the moment the RPC handler is invoked.

What was wrong

Both LowMemory RPC handlers in examples/AutoResearch/AutoResearchLowMemory.h declared multi-KB stack buffers inside the lambda body:

// line 187 (pinToggleRx)
fl::EdgeTime edges_buf[2048];  // 8 KB on the stack

// line 286 (ws2812SctTest)
fl::EdgeTime probe[1024];      // 4 KB on the stack

On LPC845-BRK (16 KB SRAM) .bss already sits at ~8.5 KB after the LowMemory build's fl::Remote + watchdog + fl::json instances. The first allocation alone pushed total SRAM demand over the 16 KB ceiling — the device crashed the moment a pinToggleRx request landed.

Verified today on hardware:

$ bash autoresearch lpc845brk --pin-toggle-rx --upload-port COM10
[lpc-pin-toggle-rx] opening COM10 @ 115200
[lpc-pin-toggle-rx] echo ok                   # device boots; echo RPC works

--- 1.1 / 1 kHz (1000 Hz / 200 ms) ---
  FAIL: no response or RPC error: None
--- 1.2 / 10 kHz ---
  FAIL: no response or RPC error: None
--- 1.3 / 100 kHz ---
  FAIL: no response or RPC error: None

Echo round-trips cleanly (handler doesn't allocate). pinToggleRx silently never replies (handler allocates 8 KB → stack overflow → trap before any reply).

The fix

Two changes to examples/AutoResearch/AutoResearchLowMemory.h:

1. pinToggleRx — buffer right-sized from 2048 to 512 entries and moved to static storage. Cap clamp updated to match. The bench cases at 1 / 10 / 100 kHz all converge sigma well before 256 period pairs, so 512 entries is comfortable.
2. ws2812SctTest — probe right-sized from 1024 to 512 entries, also static. Diagnostic-only buffer; decoder accuracy is controlled by cfg.buffer_size, not by this output probe.

Net RAM budget change:

Site	Before	After
pinToggleRx::edges_buf	8 KB stack at every call	2 KB permanent .bss
ws2812SctTest::probe	4 KB stack at every call	2 KB permanent .bss
Total	up to 12 KB stack pressure	4 KB .bss

Handlers are invoked serially over JSON-RPC, so sharing the static storage across calls is safe.

Verification

• bash lint — clean.
• Build still fits 64 KB flash (no regression on the LPC845 bloat: 8.3 KB of soft-FP / integer-div leaks past the no-soft-FP invariant on LowMemory + RX_SCT_WS2812 #3076 budget).
• On-hardware re-test of bash autoresearch lpc845brk --pin-toggle-rx --upload-port COM<N> after merge — final acceptance gate per the issue body. Expected: non-None RPC replies, with the 1 kHz case showing success=1. (Bench session needed; covered by meta(lpc845-brk): bench-validate the three on-hand-hardware checklists (deploy + #565-sidestep + SCT-RX phases) #3102 Section 1.)

Refs

• Closes lpc845: pinToggleRx + ws2812SctTest handlers stack-overflow at runtime (8 KB stack buffer on 16 KB SRAM) #3125
• Refs meta(lpc845-brk): bench-validate the three on-hand-hardware checklists (deploy + #565-sidestep + SCT-RX phases) #3102 Section 1 — bench-validation meta, this PR unblocks the Phase 1 pin-toggle gate.
• Orthogonal to cleanup(lpc): retire FASTLED_LPC_RX_SCT_WS2812 manual opt-in — derive from platform predicate instead #3128 (the FASTLED_LPC_RX_SCT_WS2812 macro retirement) — that's a build-flag cleanup, this is the runtime SRAM fix.
• Orthogonal to fix(autoresearch): gate 6 companion .cpp files on FASTLED_AUTORESEARCH_LOW_MEMORY (refs #3076) #3123 (the LowMemory .cpp companion-file gating PR) — different files, different concerns.

🤖 Generated with Claude Code

Summary by CodeRabbit

Release Notes

• Refactor
  • Optimized memory usage in low-memory configurations, reducing buffer allocations and improving efficiency on memory-constrained systems.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: e5fb2771-31cf-4892-a991-6370bd33540b

📥 Commits

Reviewing files that changed from the base of the PR and between ebfda30 and b3414a5.

📒 Files selected for processing (1)

• examples/AutoResearch/AutoResearchLowMemory.h

---

📝 Walkthrough

Walkthrough

AutoResearchLowMemory.h introduces kLowMemEdgeBufSize = 512 and kLowMemWs2812ProbeSize = 512 constants. The pinToggleRx edge capture count is clamped to 512, and its edges_buf is moved from a 2048-entry stack array to static storage. The ws2812SctTest probe buffer is similarly moved to static storage at 512 entries, down from 1024.

Changes

LPC handler buffer reduction

Layer / File(s)	Summary
pinToggleRx capture sizing and static buffer examples/AutoResearch/AutoResearchLowMemory.h	Introduces kLowMemEdgeBufSize = 512, clamps the computed edge capture count to that value (minimum 256), and changes edges_buf from a 2048-entry stack-allocated array to a static array sized to kLowMemEdgeBufSize.
ws2812SctTest static probe buffer reduction examples/AutoResearch/AutoResearchLowMemory.h	Introduces kLowMemWs2812ProbeSize = 512, changes the RX diagnostic probe from a 1024-entry stack array to a static 512-entry array, and updates the getRawEdgeTimes span accordingly.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Possibly related issues

• lpc845: pinToggleRx + ws2812SctTest handlers stack-overflow at runtime (8 KB stack buffer on 16 KB SRAM) #3125 — This PR directly implements the fix described in the issue: downsizing pinToggleRx's edges_buf from 2048 to 512 entries and moving it to static storage, and reducing ws2812SctTest's probe from 1024 to 512 entries with the same storage change.
• META: LPC845 64 KB flash bloat hunt -- clear 11 KB overage to unblock FASTLED_LPC_RX_SCT_WS2812=1 #3079 — The buffer reductions contribute to the flash/RAM budget goals for the LPC845 FASTLED_LPC_RX_SCT_WS2812=1 build referenced in this META issue.

Possibly related PRs

• FastLED/FastLED#3041: Introduced AutoResearchLowMemory.h and the original pinToggleRx/ws2812SctTest handlers whose oversized stack buffers are being reduced in this PR.

Poem

🐇 Hoppin' through the RAM so tight,
Two thousand slots? A wasteful sight!
Five twelve is plenty for each edge,
Static storage — that's my pledge.
No more stack crash in the night! ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title concisely describes the main change: moving edge buffers from stack to static storage for the pinToggleRx and ws2812SctTest handlers, and references the closed issue #3125.
Linked Issues check	✅ Passed	The PR implementation fully addresses all coding requirements from #3125: buffers moved to static storage, right-sized from 2048/1024 to 512 entries, and handler caps updated accordingly.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to addressing the stack overflow issue in #3125; no unrelated modifications are present in the changeset.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/3125-pin-toggle-stack-bomb

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}