boot-crash (plan6): module-init SIGSEGV did not reproduce on Mac¶

No source change. The reported CRASH signal=11 fault=0x38 pc=main+0x111e000 (x0=0, pc==lr) at module init could not be reproduced.

What was run (Mac, arm64, branch claude/boot-crash-2026-10-24-plan6 at 1570a78)¶

Rebuilt gfx + runtime from this checkout (make -C gfx, make -C runtime).
runtime/build/napple-runtime <Napple/work/games/<id>/<hash>> --headless --seconds N for LAS (01006BB00C6F0000) and SMO (0100000000010000):
45 s, both run concurrently: 0 CRASH lines. LAS: 2304 gfx frames, movie decoder ran (2400 movie frames, ~0.8 ms/frame), audio renderer opened (null sink). SMO: 2206 frames, 183920 draws, audio renderer opened.
3 rounds x 2 games x 2 binaries (this build and ~/Desktop/NappleLATEST/bin, main a0276af) at 10 s: 12/12 runs with 0 CRASH lines.
make -C runtime test: all suites pass (test_media 56 checks, test_timescale 21 checks, 0 failed).
cargo test --workspace: not run, no cargo on PATH in this session (~/.cargo has only a registry dir).

The only Mac-side crash logs found (~/Library/Logs/DiagnosticReports/napple-runtime-2026-10-24-1447xx..1449xx.ips) are all iOS Simulator runs (CoreSimulator paths): napple_drawable_presented unrecognised selector, a Metal texture-view validation abort in mtl_tex_view (MTLSimDriver) and archive_save -> _MTLBinaryArchive legacySerializeToURL abort. None is the SIGSEGV; iOS is out of scope. The Metal binary-archive SIGABRT is that simulator one; on the Mac the log line is binary archive: 153 of 153 pipelines loaded precompiled.
$TMPDIR/boot.log named in the brief was overwritten by this session's own run (same TMPDIR), so the original crash log is gone. The old Napple/work/logs crashes (fault=0x30 pc=main+0x775c00, fault=0x58 sdk+0x2a16c4, signal=10 main+0x107aa64) are from Oct 22-23 and different addresses.
Likely explanations, unverified: the crash was seen with a different game dir (not the two in work/games; main+0x111e000 is beyond the main+0x101fb28 reached by LAS here), a stale rewritten module set, or the iOS path. Needs the exact command/dir from the reporter.

Not reproduced, nothing fixed. No regression test added (nothing to pin).